The Conversion Window
Ken and Alice Brennan, 62 and 60, retired last month. Assets: $2.9 million traditional IRA (Ken's), $450,000 Roth, $1.1 million taxable with a $700,000 basis, and $250,000 cash. Ken's pension pays $38,000 a year beginning at 65; Social Security at 70 would pay $62,000 combined. They spend $140,000 a year. Their current-year income is unusually low because Ken's salary ended in March. From 62 to 64 they buy health insurance on the ACA marketplace and expect Medicare at 65. Their IRAs will pass to two children, 34 and 30, both in the 35% bracket. They give $25,000 a year to charity and want to continue. Alice's parents died at 91 and 94; Ken's parents died in their early 70s.
1. Should they convert to Roth in the next three years? Frame the trade-offs quantitatively and say what sets the amount.
Read the worked answer
The low-income window is a trap, not an opportunity. Three separate phase-outs sit on top of the marginal rate during ages 62-72:
- ACA premium tax credits (ages 62-64). A conversion raises MAGI and reduces the subsidy. Depending on where they sit on the schedule, the effective marginal rate on the first dollars of conversion can be 30%-60% or more on top of federal and state tax. These are the most expensive conversion dollars in the entire plan - exactly the opposite of the intuition that 'income is low, so convert.'
- IRMAA (ages 63 and later). Medicare premiums are set from MAGI two years earlier, so a conversion at 63 sets the premium at 65. Approaching a tier boundary, the last few thousand dollars of conversion can cost thousands in premiums - an effectively infinite marginal rate at the cliff.
- Social Security taxation once benefits begin at 70.
Where the value actually is: the children's 35% bracket and the 10-year distribution rule. Every dollar converted at 22%-24% that would otherwise be inherited and distributed at 35% is a 11-13 point arbitrage, before considering that the Roth has no RMDs and no tax on a decade of growth.
What sets the amount: the top of the 24% bracket, capped by the relevant IRMAA tier threshold, with the ceiling re-tested every November against that year's actual MAGI. Do not convert during 62-64 if a subsidy is in play; run those years on taxable withdrawals and start conversions at 65. Pay the tax from taxable cash - never withhold from the conversion. And never convert the charitable dollars: the $25,000 a year should come from QCDs after 70 1/2 or from the IRA at death, where charity pays no tax at all.
2. Sequence the withdrawals from 62 to 73 and justify each step.
Read the worked answer
62-64: spend cash and taxable assets. Keep MAGI below the ACA subsidy threshold. Sell the highest-basis lots - lowest gain, lowest tax. Justification: every dollar of IRA income here costs more than the tax rate, because it drags the premium credit with it.
65-69: pension starts at 65; begin conversions. Spend taxable assets to pay the conversion tax. Fill to the top of the 24% bracket, staying under the IRMAA tier. Justification: the subsidy distortion is gone; the IRMAA ceiling is a visible, testable line.
70: claim Social Security. Delaying to 70 is the best inflation-indexed annuity they can buy, and the family history points to a long joint life - Alice's parents reached 91 and 94. The joint-and-survivor benefit is worth more than the withdrawals it costs. Justification: longevity insurance should be purchased with the highest-mortality-risk asset in the plan, and their human capital is the only thing that does not need to be diversified.
70 1/2 onward: QCDs of $25,000. Satisfies the charitable goal, reduces MAGI, and avoids the itemized-deduction limit. Justification: the cheapest charitable dollar is one that never enters income.
72-73: fill the remaining bracket before RMDs begin. Convert the remainder up to the ceiling.
The asymmetry that should push them harder than they expect: the widow's penalty. When one dies, the survivor files single on the same IRA. Converting at joint 24% rates while both are alive is materially cheaper than the survivor converting at single 32%-35% rates later. That is the strongest argument for using the full bracket, not the weakest.
3. Ken wants to leave the taxable account untouched 'for the kids' and fund spending from the IRA instead. What is wrong with that, and how do you show him?
Read the worked answer
It is exactly backwards, and the arithmetic is simple enough to put on a napkin.
To net $100 of spending:
- From the IRA: $100 gross, taxed at 24% = $76 net. It also raises MAGI, which taxes more of Social Security later and can trip IRMAA.
- From the taxable account, selling a lot with a 30% embedded gain: $100 of proceeds, $30 of gain, tax at 28.8% = $8.64, $91.36 net.
The same dollar of spending costs about 20% more when it comes out of the IRA.
The inheritance logic is inverted too. What the children inherit:
| Asset | What the children receive |
|---|---|
| Taxable account | Step-up in basis - no income tax, ever |
| Roth IRA | Tax-free, with 10 years of further growth |
| Traditional IRA | Fully taxable, forced out within 10 years, at their 35% bracket |
The account Ken wants to protect is the best asset to inherit. The account he wants to spend is the worst. Spending the IRA first and leaving the taxable account intact transfers wealth to the government at 35%.
The nuance that keeps this honest: within the taxable account, spend the high-basis lots and hold the low-basis lots. Low-basis shares get the full benefit of the step-up; high-basis shares get almost none, so they are the ones to sell. With a $400,000 embedded gain on $1.1 million, that distinction is worth real money.
The one-page version for Ken: "You have three buckets. One is taxed at 35% when the kids touch it. One is never taxed. One is taxed at 0% if you die holding it. Spend the one you were planning to save, and save the one you were planning to spend."