[TX-02] Roth Conversion Strategy

Filling brackets, the age-63 wall, and the widow(er)’s trap — converting to smooth a lifetime rate, not to cut this year’s tax

Pillar: Tax Optimization & Decumulation · Applies to: Anyone with a large traditional balance heading toward RMDs — especially married couples and anyone past the ER-02 ladder years Last verified: August 2026 · Refresh cadence: Annual (brackets and thresholds via FN-02) + event-driven (senior-deduction sunset after 2028) Related: ER-02 Roth Ladder · TX-01 0% Capital Gains · TX-03 Withdrawal Sequencing · TX-04 RMDs & QCDs · HC-06 IRMAA · SS-01 Claiming Age · SS-03 Taxation of Benefits · FN-02 Key Numbers

Not advice. Conversions are irreversible — recharacterization was abolished in 2018 — and they interact with Medicare surcharges, Social Security taxation, health subsidies, and your heirs’ brackets. The arithmetic here uses verified 2026 figures, but the answer depends on projections about your own future that no article can make. Model your numbers.

---

TL;DR

• This article is about rate arbitrage; ER-02 is about access. The ladder exists to get money out before 59½ without a penalty. Conversion strategy exists to move dollars from a year when they’d be taxed high to a year when they’d be taxed low. Same transaction, different reason — and the sizing rules are different.
• The only question that matters: what rate would this dollar face now, versus what rate would the same dollar face when it eventually comes out? If later is higher, convert. Everything below is machinery for estimating “later” honestly.
• “Later” is usually higher than people model, for four reasons most projections omit: RMDs stack on top of Social Security, the survivor files single at halved brackets, IRMAA is priced two years back, and heirs drain inherited traditional accounts inside ten years at their own peak-career rates.
• The widow(er)’s trap is the single largest under-modelled cost in retirement tax planning. The worked example in §3 shows a household whose income falls 12% at the first death while its combined tax and Medicare bill rises 35%.
• Convert hardest at 60–62 — after the paycheck stops, before the age-63 IRMAA lookback opens (HC-06 §6). That airspace never comes back.
• The honest counter-case is in §7. Charitable intent, a genuinely low future rate, or paying the tax from the IRA itself can each make converting the wrong move.

---

1. The Only Question

A traditional IRA is a partnership with the IRS in which the government owns an undetermined share. A conversion buys them out at today’s price.

Convert when:  your marginal rate today  <  the marginal rate that dollar
                                             would face when it comes out

That’s it. Everything else — brackets, IRMAA, the survivor, the heirs — is machinery for estimating the right-hand side. And the right-hand side is not “my rate in retirement,” which is the comparison most people make. It is the rate on the specific dollar, in the specific year it is forced out, under the filing status that will apply then.

Four things make the right-hand side higher than a naive projection:

Force	Why it raises the future rate
RMDs are non-optional and they stack	At 73 or 75 (FN-02) a percentage of the balance comes out whether you need it or not, on top of Social Security and pensions — filling the low brackets you’d hoped to convert into
The survivor files single	Halved brackets, halved IRMAA thresholds, one Social Security check gone (§3)
IRMAA is priced two years back	A conversion at 63 sets a Medicare premium at 65 (HC-06) — and IRMAA is a cliff, not a slope
Heirs drain in ten years	Inherited traditional accounts land on children in their 50s, at their career-peak rates (TX-03 §5)

Deferral felt free while you were accumulating. It was a loan at a rate nobody quoted.

2. The 2026 Brackets, and What “Filling” Means

Verified against Rev. Proc. 2025-32 §3.01 — taxable income, after deductions: ✅

Rate	Married filing jointly	Single	Head of household
10%	$0 – $24,800	$0 – $12,400	$0 – $17,700
12%	– $100,800	– $50,400	– $67,450
22%	– $211,400	– $105,700	– $105,700
24%	– $403,550	– $201,775	– $201,750
32%	– $512,450	– $256,225	– $256,200
35%	– $768,700	– $640,600	– $640,600
37%	above	above	above

Note the two rows in bold. The 22% and 24% bands are where nearly all conversion decisions live, and note how much narrower they are for a single filer — the 22% band tops out at $105,700 single versus $211,400 joint. That asymmetry is the widow(er)’s trap, expressed as a table.

“Filling a bracket” means: take your existing taxable income, subtract it from the top of your chosen bracket, and convert the difference. With a 2026 standard deduction of $32,200 MFJ / $16,100 single (plus $1,650 per married spouse or $2,050 unmarried at 65+) ✅, a couple with $150,500 of taxable income has $60,900 of unused 22% space before touching the 24% bracket. Left unused, that space expires on December 31 and never returns.

3. The Widow(er)’s Trap

This is the section that changes plans.

At the first death the household loses the smaller Social Security benefit (SS-01 §4), keeps essentially all of the portfolio and its RMDs, and moves from joint brackets to single brackets — halved. The result is routinely a falling income and a rising tax bill.

Worked example

Couple, both 75, 2026 figures. Income: $130,000 of RMDs plus $66,000 of Social Security (85% taxable = $56,100). AGI $186,100, call it $186,000. Both 65+, so the standard deduction is $32,200 + $3,300 = $35,500.

(The 2025–2028 senior deduction is set aside here for clarity — at this income it is mostly phased out anyway, and it expires after 2028. Its effect is §5.)

	Both alive (MFJ)	Survivor (single)
RMDs	$130,000	$130,000
Social Security (taxable portion)	$56,100	$34,000 (one check lost)
AGI	$186,000	$164,000 — down 12%
Standard deduction	$35,500	$18,150
Taxable income	$150,500	$145,850
Marginal bracket	22%	24%
Federal income tax	$11,600 + 22% × $49,700 = $22,534	$17,966 + 24% × $40,150 = $27,602
IRMAA tier (on this MAGI)	≤ $218,000 → $0	$137k–$171k → $2,884.80
Total federal cost	$22,534	$30,487

Income fell 12%. The bill rose 35%. The survivor pays $7,953 more per year, forever, on less money — from a household event that changed nothing about the portfolio.

Three forces compounded: the brackets halved, the standard deduction halved, and IRMAA went from zero to tier 2 because the single thresholds are half the joint ones (HC-06 §4).

What conversions do about it. Every dollar converted while both spouses are alive is a dollar taxed at 22% instead of 24%-plus-IRMAA later, and a dollar removed from the RMD base that creates the problem. That couple has $60,900 of unused 22% space each year (§2). Using it for seven years moves $426,300 into Roth at 22% — shrinking the RMD stream the survivor inherits, and doing it at a 2-point discount before the IRMAA cliff is even counted.

The planning rule: model every conversion decision against single brackets at the older spouse’s mortality age, not against joint brackets forever. A plan that only looks right while both spouses are alive is a plan that fails at exactly the moment it’s needed.

4. The Age-63 Wall

Medicare’s IRMAA surcharge uses a two-year lookback, so the year you turn 63 prices your first Medicare premium year (HC-06 §6). The conversion window has a shape:

Ages	Status	What to do
Retirement – 62	The free window. Pre-lookback; no IRMAA consequence at all	Convert hardest here. If still on an ACA plan, weigh against the subsidy (HC-01 §5)
63 – 64	The wall goes up. Conversions now carry an IRMAA toll two years out	Either stay under a threshold with a $3,000–$5,000 buffer, or cross decisively and fill the tier
65 – 72	Managed. Each year’s MAGI prices a premium year	Keep a running target; confirm in December
73/75+	RMDs arrive and are non-optional MAGI	Whatever you failed to convert now sets a permanent floor under your IRMAA tier

Do not let IRMAA veto a good conversion. HC-06 §3 prices it precisely: crossing one tier costs a couple $2,296.80 for a single year, because only one tax year feeds each premium year. Against a conversion that saves 2+ points across decades of RMDs and a survivor’s compressed brackets, a one-year toll is a rounding error. The mistake runs both ways, though — crossing a tier to convert $1,200 is indefensible. Once you decide to cross, fill the tier; the toll is identical at $1 over or $27,000 over.

5. The Senior-Deduction Band (2025–2028)

OBBBA created an extra $6,000 deduction per person 65+ for tax years 2025–2028. It phases out at 6 cents per dollar of MAGI above $75,000 single / $150,000 joint, disappearing entirely around $175,000 single / $250,000 joint ✅ — which implies the reduction runs at 6% per eligible person, so 12% combined for a couple where both are 65+ ◻️.

The marginal effect, computed properly. Inside the band, each extra dollar of MAGI also destroys 6 cents (single) or 12 cents (couple, both 65+) of deduction, so taxable income rises by more than a dollar:

In the phase-out band	Taxable income rises per $1 of MAGI	22% bracket becomes	24% bracket becomes
Single, 65+	$1.06	23.3%	25.4%
MFJ, both 65+	$1.12 ◻️	24.6%	26.9%

A correction to two other pages in this wiki. ER-02 §4 and TX-03 §4 each describe this as “an effective ~6% surtax.” That overstates it by roughly four times: 6% is the phase-out rate, not the surtax. The real uplift is your marginal rate times 1.06 or 1.12 — about 1.3 to 2.9 percentage points. Worth knowing about and sizing around; not worth distorting a conversion plan for.

The part that does deserve planning: it expires. From 2029 a 65+ couple’s deductions drop by up to $12,000, so the same gross income produces more taxable income. Conversion capacity contracts in 2029. If you are 65+ now, the 2026–2028 window is temporarily wider than the one that follows it → FN-03 §5.

6. Sizing: the Ceilings, in Ascending Order

Each year, pick the lowest ceiling that binds, and convert to it with a buffer:

#	Ceiling	2026 value (MFJ)	Binds when
1	Standard deduction	$32,200 (+$3,300 at 65+)	Always — never file with an empty deduction (TX-03)
2	Top of the 12% bracket	$100,800 taxable	The classic “fat rung” ceiling
3	0% capital-gains collision	$98,900 taxable	You’re also harvesting gains — a 12% conversion really costs 27% (TX-01 §5)
4	ACA subsidy lines	FPL-based	Pre-65 and subsidized (HC-01)
5	Top of the 22% bracket	$211,400 taxable	The workhorse ceiling for large balances
6	IRMAA tier	$218,000 MAGI (first tier)	Age 63+
7	Senior-deduction band	$150,000 MAGI	65+, through 2028 (§5)
8	NIIT	$250,000 MAGI	Rarely binding on conversions alone — conversions aren’t investment income, but they raise MAGI ◻️
9	Top of the 24% bracket	$403,550 taxable	Very large balances with a survivor problem

Note that ceilings 3, 4, 6, 7 and 8 are MAGI-based while 1, 2, 5 and 9 are taxable-income-based. Mixing them up is the most common sizing error in this whole subject — MAGI sits above the standard deduction, so a couple with $150,500 of taxable income has $186,000 of MAGI, and it’s the larger number that IRMAA reads.

7. When Not to Convert

The honest counter-cases. Each one is a real reason to stop:

• Your future rate is genuinely lower. A modest traditional balance, no pension, and delayed Social Security can produce RMDs that never leave the 12% bracket. Converting at 22% to withdraw at 12% destroys value. Project the RMDs before assuming.
• You are charitable. A QCD at 70½+ moves pre-tax dollars to charity at a 0% rate (TX-04) — better than any conversion. And naming a charity as the beneficiary of the traditional IRA passes it at full value. Convert what the kids will inherit; leave the charity’s share pre-tax. Converting first and donating later wastes the best feature of the account.
• You’d pay the tax from the IRA itself. Paying conversion tax with converted dollars shrinks the amount that lands in Roth and — under 59½ — triggers a 10% penalty on the withheld portion (ER-02 §5). The arbitrage only works cleanly when outside cash pays the bill.
• Your heirs are in low brackets, or are charities. The ten-year-drain argument reverses.
• A move to a no-income-tax state is coming. Wait. The same conversion can cost 0% or 10%+ depending on which side of the moving truck it lands (State notes, ST-03).
• You may need Medicaid long-term care within the lookback. Conversions accelerate income into the wrong years; that is a different planning problem (HC-08, ST-01 dimension 4).
• You’re under 59½ and will need the money inside five years. That’s an ER-02 clock problem, not a rate problem.

8. Three Scenarios

A. The couple who under-converted (both 63, $1.9M traditional). Their advisor’s spreadsheet says “avoid IRMAA,” so they cap MAGI at $217,000 every year — converting roughly $60,000/year less than filling the 22% bracket would allow. By 75 their RMDs alone put them in tier 2 permanently, and the survivor faces §3’s compression on top. Accepting tier 1 ($2,296.80/year for the couple) to convert an extra $60,000 annually adds 3.8 points of effective rate in the years they cross (HC-06 §7). They were optimizing the small number against the big one — a one-year toll against a permanent floor.

B. The couple who should stop (both 68, $450k traditional, $52k spending). Projected RMDs at 75 are roughly $18,000/year on top of $48,000 of Social Security. Their taxable income never leaves the 12% bracket in any year they model, including the survivor’s. Converting at 22% today to avoid 12% tomorrow is value destruction. They convert only to the top of the 12% bracket, and only in years they aren’t harvesting gains (TX-01 §5). The correct answer here is less conversion, and it takes a projection to see it.

C. The charitable widower (77, $1.4M traditional, gives $25k/year). Conversion looks attractive on bracket math alone — he’s single, in the 24% band, with the trap already sprung. But he is giving $25,000/year regardless, and QCDs move that money at 0% while satisfying the RMD (TX-04). He routes giving through QCDs first, converts only what exceeds his charitable capacity, and names the remaining traditional balance to charity at death, leaving the Roth to his children. The pre-tax account is the best asset a charity can receive and the worst one an heir can.

---

💡 Pro-Tips

1. Project RMDs on a “do nothing” path first. You cannot size a conversion program without knowing the bomb you’re defusing — and sometimes the projection says there isn’t one (§7).
2. Model the survivor explicitly, at the older spouse’s mortality age, in single brackets. It is the highest-value ten minutes in retirement tax planning.
3. Convert hardest at 60–62, before the age-63 lookback opens. That window doesn’t come back.
4. Once you decide to cross an IRMAA tier, fill it. The toll is fixed; the marginal cost of the next $20,000 inside the tier is zero.
5. Never withhold tax from the conversion itself if under 59½, and prefer outside cash at any age.
6. Decide each year whether it’s a conversion year or a harvest year — they compete for the same bracket space, and doing both costs 27% (TX-01 §5).
7. Convert in December, after fund distributions post and income is known, with a $3,000–$5,000 buffer under any MAGI cliff.
8. If you’re charitable, do QCDs before conversions at 70½+. Pre-tax dollars to charity at 0% beat any rate arbitrage you can construct.

⚠️ Common Pitfalls

1. Comparing “my rate now” to “my average rate in retirement” instead of the marginal rate on the specific dollar in the year it’s forced out.
2. Modelling only the joint-filing years and never running single brackets — the trap in §3.
3. Confusing MAGI ceilings with taxable-income ceilings. MAGI sits above the standard deduction; IRMAA and the ACA read the bigger number.
4. Letting a one-year IRMAA toll veto a decades-long arbitrage — or crossing a tier for a trivial conversion.
5. Converting at 22% when your projected RMDs never leave 12% — the mistake §7 exists to prevent.
6. Converting the dollars you were going to give to charity anyway, forfeiting the QCD and the charitable beneficiary designation.
7. Paying the tax out of the converted amount, shrinking the transfer and — under 59½ — adding a penalty.
8. Treating the 2026–2028 senior deduction as permanent. It expires; conversion capacity contracts in 2029.
9. Believing the phase-out band adds a 6% surtax. It adds roughly 1.3–2.9 points (§5).
10. Converting immediately before a move to a no-income-tax state, paying a state bill you were weeks away from avoiding.
11. Assuming rate permanence is rate certainty. OBBBA removed the scheduled sunset, not Congress’s ability to legislate. Convert on your own bracket math.

✅ Actionable Checklist

Once (design)

• Project RMDs at 73/75 on a do-nothing path; identify the marginal bracket they land in
• Run the same projection under single brackets at the older spouse’s mortality age
• Decide the estate map: charity ← traditional · heirs ← Roth · step-up ← taxable (TX-03 §5)
• Write down the target ceiling for each life phase (§6)

Ages 60–62

• Convert aggressively — no IRMAA consequence yet
• If on an ACA plan, price the subsidy cost per converted dollar first (HC-01 §5)

Age 63 onward

• Recognize this year prices your age-65 Medicare premium (HC-06)
• Choose a threshold and hold a $3,000–$5,000 buffer
• 65+ through 2028: check whether you’re in the senior-deduction band (§5)

Every December

• Tally actual income including fund distributions
• Confirm this year’s binding ceiling from FN-02 — the lowest one that applies
• Decide: conversion year or harvest year (TX-01)
• Convert with zero withholding; pay from outside cash; log the amount and file Form 8606

---

State notes (→ ST-01, ST-04): A conversion is ordinary income in most states that have an income tax, so the true cost is the federal rate plus your state’s marginal rate — and that can move the answer by several points. In the nine no-income-tax states the entire program runs federal-only (ST-04 §1.1), which is why the SS-01 bridge-to-70 drawdown and the conversion program are so much cheaper there. Three states — Illinois, Pennsylvania, and Mississippi — tax wages but exempt essentially all retirement income ✅, which can make conversions state-free there too, though the treatment of a conversion specifically differs from the treatment of a distribution in some of them ◻️ — confirm with the state’s revenue department before sizing around it. Sequencing around a move dominates everything else on this page: convert after establishing domicile in a low-tax state and before moving into a high-tax one, and read ST-03 on making the domicile change survive an audit, because sticky states audit departures. Finally, note the interaction with state estate tax: a large Roth is still in your taxable estate, so in a low-exemption state like Oregon ($1M) or Washington ($3M, frozen) converting reduces income tax without reducing estate exposure (ST-01 dimension 3, EP-05).

Sources & further reading (verified Aug 2026)

• Rev. Proc. 2025-32 §3.01 — 2026 tax rate tables for all filing statuses (read directly; the MFJ 22% band runs to $211,400 and the single 22% band to $105,700) and §3.15 standard deductions
• IRC §408A (Roth rules); Form 8606 instructions (reporting conversions and basis); IRS Pub. 590-A/590-B
• OBBBA (P.L. 119-21) — bracket permanence; the 2025–2028 senior deduction, its 6%-per-person phase-out above $75,000/$150,000 MAGI, and its expiry after 2028
• 20 CFR 418.1205 and CMS 2026 premium releases for the IRMAA thresholds used in §3 and §4 — full table in FN-02, mechanics in HC-06
• ER-02 for the pre-59½ ladder and the two five-year clocks; TX-01 §5 for the conversion-versus-harvest collision; TX-04 for QCDs; SS-01 §4 for the survivor benefit mechanics behind §3