Nine small years and one enormous one. The heir's rate was the smaller half of the argument.

This article first appeared in The Nest Egg, my daily email, on 2026-10-07. Read the original issue. It is educational, not a recommendation for you. The household in it is a hypothetical example, and market figures are as of the date shown and will have changed.

Every figure below names its document. Every calculation is shown. Nothing here is a forecast.

Before anything else: there was no letter yesterday. The publishing account was signed out and the issue could not be sent. Nothing was wrong with the arithmetic. What follows is the letter that should have arrived on Tuesday morning, with Tuesday’s close in place of Monday’s.

Good morning. A narrow advance: the S&P, the Dow and the Nasdaq 100 each added about half a percent while the Russell 2000 went the other way, and the VIX slipped to 15.01. Then the person every conversion in this letter has been for, described at last.

In today's issue

The tape. Three indexes higher, the Russell 2000 the exception, and volatility down again. Two minutes.

28.9. The denominator the ten-year rule throws away, and the year it leaves behind. Five minutes.

$52,807.95. What converting is worth once the heir is a real person instead of a rate. Four minutes.

One more thing. Where the conversion stops paying, and the one word that changes the whole answer. Two minutes.

The tape

Close of October 6, 2026. Index and futures levels from CNBC US Markets, pulled at 4:15pm ET. Treasury yields are as of that moment, not the official close. Gold is the December 2026 contract and WTI the November 2026 contract, both unchanged from Monday.

A narrow advance. The S&P, the Dow and the Nasdaq 100 all gained roughly half a percent, and the VIX fell 3.29% to 15.01. The Russell 2000 was the lone decliner at 0.59%, which is the first time in a week the small-cap index has parted company with the other three. Crude steadied at $89.73 on the November contract, which expires on October 20, and gold added 0.16% to $4,193.80. Both ends of the curve rose a little, the 10-year by 1.5 basis points to 5.286% and the 30-year by 2.0 to 5.661%.

The story

Monday's letter ended somewhere uncomfortable. With a charitable distribution running every year, the survivor's Medicare surcharge is zero, and a conversion cannot improve on zero. So the entire case for converting collapsed onto one number that belongs to someone else: a child's marginal rate, with a break-even of 16.42%.

In twenty sessions this letter has never once described that child.

Today it does, and the answer is not the one the break-even suggested. The rate turns out to be the smaller half of the argument.

The position

The child is 48 in 2026, married, filing jointly, with $180,000 of household income. The survivor dies in 2036, when the child is 58.

One input needed checking before anything else, and it is the kind that has caught this letter four times. The household's standard deduction has been $35,500 all month. The 2026 joint standard deduction is $32,200. The difference is two additional deductions of $1,650 for being 65 or older, which both parents get and the child does not. The household figure was right. The child's is $32,200, and using $35,500 for a 48-year-old would have been wrong by $3,300 in every year of the model.

So the child's taxable income before inheriting anything is $147,800, which sits in the 22% bracket with $63,600 of room before the 24% bracket at $211,400.

The arithmetic

Step one. The child is not an eligible designated beneficiary.

That status is what buys a beneficiary the right to stretch an inherited account over a lifetime, and the IRS required minimum distribution FAQ lists exactly who qualifies: a surviving spouse, a child who has not reached the age of majority, a disabled person, a chronically ill person, and a person not more than ten years younger than the owner.

A 58-year-old adult child of a parent who died at 83 is none of those. So the ten-year rule applies, and Publication 590-B is blunt about what that means: distributions to a designated beneficiary who is not an eligible designated beneficiary must be completed within ten years of the death of the owner.

Step two. Whether anything is required before year ten, which is the hinge.

This is the input the whole letter turns on, and Publication 590-B never states it in a single sentence. It states two sentences that together settle it.

The first, under Payment under the 10-year rule: "If the IRA owner dies before the required beginning date and the 10-year rule applies, no distribution is required for any year before the 10th year."

The second, under Owner Died on or After Required Beginning Date: "If the owner died on or after their required beginning date and you are a designated beneficiary, base your required minimum distributions for years after the year of the owner's death on the longer of: your single life expectancy shown in Table I in Appendix B; or the owner's life expectancy."

The relief is conditioned on dying before the required beginning date. The survivor dies at 83, long past hers. So her child gets no relief: a distribution is required every year, and the account still has to be empty by year ten.

I am showing the two sentences rather than asserting the rule because this letter has been wrong on four inputs in nine sessions and this is exactly the shape of all four.

Step three. What that schedule actually produces.

Table I gives a 58-year-old a single life expectancy of 28.9 years, and the denominator falls by one each year after.

The $165,000 slice we are deciding about grows at 5% for the ten years to her death and arrives as $268,767.61.

Divide that by 28.9 and the first required distribution is $9,299.92. By year nine it has crawled to $13,740.21. Nine years of required minimums take out $102,546.14 while the account keeps growing, and then the ten-year rule arrives and demands everything left.

Year ten is $287,101.76, in one year, on top of the child's own income.

What a 58-year-old child must withdraw from an inherited traditional IRA when the owner died after their required beginning date. Table I and the ten-year rule from IRS Publication 590-B, read October 5, 2026. 2026 brackets and the $32,200 joint standard deduction from the IRS tax-year-2026 inflation release.

Step four. What that one year costs.

Their taxable income in year ten is $434,901.76. The 2026 joint 32% bracket begins at $403,550, so the sweep carries them through 24% and into 32%.

Tax on the sweep alone: $70,140.56, a rate of 24.43% on that single distribution.

Across all ten years the child distributes $389,647.90 and pays $92,700.71, a blended rate of 23.79%.

A 28.9-year denominator attached to a ten-year deadline is not a schedule. It is a deferral with a cliff at the end, and the cliff is where the tax is.

Step five. The Roth is a different rule, not just a different rate.

Here is the sentence almost nobody quotes, from Publication 590-B: "If a Roth IRA owner dies, the minimum distribution rules that apply to traditional IRAs apply to Roth IRAs as though the Roth IRA owner died before their required beginning date."

Read that against step two. A Roth owner is always treated as having died before the required beginning date, because a Roth has no lifetime distributions at all. So the first sentence applies, and nothing is required before year ten.

The child who inherits the traditional account must take something every year, in whatever year it lands. The child who inherits the Roth takes nothing until year ten and may take it in any pattern they like, in any year they like, tax free.

The conversion does not only change what the money costs. It changes who controls when it moves.

The break-even, corrected

Monday's 16.42% is right, and it is right for a narrower reason than it looked.

If both accounts hold the same money, grow at the same rate, and are drained on the same schedule, then the Roth leaves the family the slice times growth times (1 minus 16.42%), and the traditional leaves it the slice times growth times (1 minus the heir's rate). The growth term is on both sides and cancels completely.

So the break-even is the rate, and only the rate, and the horizon is irrelevant. That is the opposite of the usual claim that a long runway favours converting. A long runway makes the gap bigger in dollars. It does not change which side wins.

But the two accounts are not drained on the same schedule, because step five says they cannot be. That difference does grow with time, and it has to be priced separately.

Valuing everything at the end of the child's tenth year, twenty years out, with distributed money reinvested at 3.90% after tax:

The Roth, taken whole in year ten: $365,888.69, after subtracting the $27,100.40 conversion tax carried forward at 5%.

The traditional on the required schedule: $313,080.74.

Converting is worth $52,807.95. The 5.58-point rate advantage explains roughly half of it. The schedule explains the rest.

What the four of them would say

The portfolio manager. Look at what happens when the child does the obviously sensible thing. Levelling the withdrawals to stay inside the 22% bracket cuts the blended rate from 23.79% to 22.00%, and leaves them $4,893.29 poorer, because every dollar pulled out early stops compounding at 5% and starts compounding at 3.90%. The tax-optimal schedule and the wealth-optimal schedule point in opposite directions here. Optimising the rate is not the same as optimising the money, and almost every piece of advice written about inherited accounts optimises the rate.

The CPA. The 28.9 denominator is the thing to understand. It is calculated for a payout over a 58-year-old's remaining life, and then a ten-year deadline amputates it at year ten. The required minimums are sized for a schedule that is not allowed to finish, which is why they remove so little and why the final year is nearly three times the previous nine combined. Any beneficiary who reads "required minimum distribution" and assumes the minimum is a plan has already lost the argument.

The financial planner. The honest version of today is that the decision was never about the parents. The conversion is priced against a rate and a schedule belonging to a 48-year-old whose income in 2037 nobody knows. That is not a reason to avoid the decision. It is a reason to stop calling it a tax calculation and start calling it what it is, which is a bet on a person.

The estate attorney. And it is a bet you can hedge, which nothing above does. Beneficiary designations are per account and can be split. A household holding both a traditional IRA and a Roth can name different beneficiaries for each, or the same beneficiaries in different proportions, and it costs nothing to do and nothing to change. The arithmetic in this letter assumes one child inheriting one slice. Real households almost never look like that, and the form that fixes it is the shortest one in the file.

One more thing

Everything above said convert. Here is where it reverses.

Hold the inheritance constant and move only the child's own income, keeping the levelled schedule:

At $60,000 of other income, the blended rate on the inheritance is 12.00%. At $90,000, still 12.00%. At $120,000 it is 18.03%. At $180,000, 22.00%. At $260,000, 24.00%.

The break-even is 16.42%. So somewhere between $90,000 and $120,000 of the child's household income, the entire argument flips, and converting at 16.42% destroys value rather than creating it.

That is a narrow band, and the parents are standing on the wrong side of it if the child is a teacher, or works part time, or retires early, or is between jobs in 2037. The conversion is not a strategy that is right or wrong in general. It is a bet that one specific person will still be earning well in a year that is eleven years away.

And one word changes all of it: where. Every figure in this letter is federal. The child's state of residence adds its own rate to the traditional path and nothing at all to the Roth.

Tomorrow

So tomorrow we add the state.

A child in a state with no income tax and a child in a high-rate state are two different decisions on the same account, and the difference can be larger than the 5.58-point federal rate advantage that carried today's answer.

We will price both ends, name the rates from each state's own revenue department, and find where the conversion flips on geography alone.