The ACA subsidy cliff at 58: one IRA-heavy couple, three plans

Plan Mirror · Updated 2026-11-04

Retire before 65 and you need health insurance until Medicare. For many early retirees that means the ACA marketplace, and a premium tax credit that depends on one number: the household's income for the year. Most explanations of the "subsidy cliff" stop at the rule: go over 400% of the poverty line and the credit disappears. For a couple whose savings sit mostly in traditional IRAs, that's half the problem. Every IRA dollar they take is income, so they can land above the cliff. But in the years they live on other savings they can have too little income, and below 100% of the poverty line there's no credit either.

We ran one couple through three ways of drawing their money in those years. Depending on the plan, they pay anywhere from $110,553 to $247,461 in premiums over seven years. The odds that their savings last change far less.

The household

A married couple in Texas, which has no state income tax, both 58, retiring this year and planning to 95. That's 38 years of retirement, through 2063. They buy marketplace coverage from 2026 until they turn 65, through 2032.

Household

A couple, both 58

Retires

At 58, in 2026

Plans through

Age 95 (2063)

Invested

$1,600,000: $200,000 taxable brokerage, $1,350,000 traditional 401(k)/IRA, $50,000 Roth

Spends

$70,000 a year in today's dollars, taxes and healthcare included

Social Security

$40,000 a year combined, in today's dollars, from 67

401(k)/IRA before 59½

Not touched until 59½

Assumes

6% average return, 2.5% inflation

Taxes

Married filing jointly, Texas (2026 rules)

Most of their savings, $1,350,000, is in traditional IRAs. A brokerage account holds $200,000, about three years of spending, and a Roth IRA $50,000. Social Security starts at 67.

The spending figure is all-in. It covers living costs, income tax and the marketplace premium. Plan Mirror pays all of them out of the same $70,000, so every dollar of premium the credit doesn't cover is a dollar the couple doesn't get to spend on anything else.

How we priced the insurance

Plan Mirror doesn't have a quote for this couple's county, so we priced their coverage from national figures. The starting point is the average benchmark premium (the second-lowest-cost silver plan) for a 40-year-old in 2026: $625 a month, KFF's national average. Premiums rise with age on the federal default age curve, so for two 58-year-olds the benchmark comes to $29,906 for the year, and by 64 it's $40,834 (later years also grow with inflation).

The credit is the benchmark premium minus an expected contribution, a share of the household's income that rises as income rises. It applies between 100% and 400% of the poverty line. For a couple buying 2026 coverage the poverty line is $21,150, so the window runs from $21,150 to $84,600 of income. Income here means modified adjusted gross income: traditional IRA withdrawals and Roth conversions count in full, and so do capital gains, but the part of a brokerage sale that's the couple's own money coming back doesn't count, and nor does a Roth withdrawal. For this household, Plan Mirror estimates that 20% of each brokerage sale is gain. You can change that by entering the account's cost basis. We assumed they buy the benchmark plan itself.

The three plans

Plan Mirror's default order. Spend the brokerage account first, then the IRAs, with no Roth conversions. This is the plan on the charts below.

Convert while living on the brokerage account. The same order, plus a Roth conversion of $40,000 a year (in today's dollars) from 2026 through 2028, the years they'd otherwise have almost no income.

Fill the 12% bracket. A common Roth conversion rule of thumb: every year, convert enough to fill the 12% federal tax bracket.

Answer one: what each plan does to the credit

The default order falls through the floor. For the first three years the couple lives almost entirely on the brokerage account. Most of each sale is their own money coming back, so their income in 2026 is just $14,000 of gains: 66.2% of the poverty line. That's below 100%, so they get no credit, and they pay the full $29,906 benchmark premium out of their $70,000. The same happens in 2027, and in 2028, when a small IRA withdrawal of $3,531 tops up the last of the brokerage money: income is still only 78.9% of the poverty line.

In 2029, at 61, the brokerage account is spent and the IRAs pay the bills. Their income jumps to $75,382, 331% of the poverty line, and the credit covers $28,009 of the $35,517 premium. They pay $7,508. Same couple, same spending, same insurance: the only change is where the money comes from.

Over the seven marketplace years the credit totals $121,597, and they pay $125,865 in premiums.

Converting early fills the gap, and opens a new one later. A $40,000 conversion lifts their 2026 income to $54,156, 256.1% of the poverty line. The credit is now $25,236 and they pay $4,670, at the cost of $780 in income tax. In 2027 and 2028 it works the same way.

Then the five-year rule catches up. Plan Mirror lets a converted amount come out of the Roth tax-free once it has waited five years, and it spends that money before the IRAs. So the couple's income drops back to $22,614 in 2031 and $23,179 in 2032, 94.5% of the poverty line both years, and the credit is gone again: they pay the full $39,201 at 63. Over the seven years the credit totals $136,909, more than the default order, and premiums $110,553. The gain is real but smaller than the first year suggests, because the early fix moves the low-income problem to the end of the marketplace years.

Filling the 12% bracket goes over the cliff, every year. In 2026 the conversion is $133,000 and their income $149,825, 708.4% of the poverty line. Above 400% there's no credit at all, and that's true in every marketplace year. They pay the whole benchmark premium, $247,461 over seven years, and $14,124 in income tax in the first year alone. A conversion plan built for the tax brackets, with no regard to the ACA, costs this couple their entire credit.

Answer two: steady returns

Plan Mirror's first test grows the portfolio at 6% a year, raises spending with 2.5% inflation and works out each year's taxes. On the default order the couple finishes at 95 with $5.7 million, about $2.3 million in today's dollars.

The early-conversion plan ends with $2.3 million in today's dollars, and the bracket-filling plan with $2 million. That gap isn't the insurance. When a plan converts to Roth, Plan Mirror keeps the couple's spending after tax the same as on the default order and pays the conversion tax on top, from the portfolio. Filling the 12% bracket every year means a lot of tax paid early.

Dollar figures in the table are in each year's own dollars.

Two quick stress tests

On the default order, with spending 10% higher the money still lasts, and they end with about $1.7 million in today's dollars. With returns 2 points lower it lasts too, with $769,000 left. The two conversion plans also last in both stress tests.

Answer three: a century of real markets

The historical replay runs the plan through actual U.S. stock returns and inflation from every start date since 1926 Q3 with a full 38-year window behind it.

On the default order, 248 of 249 start dates last to 95: 99.6%. Early conversions give 248, and filling the 12% bracket 237, again because of the extra conversion tax.

Answer four: 1,000 possible futures

The Monte Carlo test generates 1,000 random market paths. The typical one grows at the same 6% a year as the steady projection, with returns swinging around it at an assumed 15% volatility a year.

On the default order 924 of 1,000 paths last: 92.4%. Early conversions give 92.7% and filling the 12% bracket 89.6%.

All three side by side

Why don't the insurance differences show up here? Because the tests measure whether the portfolio lasts, and the couple's spending is all-in. The credit they lose comes out of what they live on in their late fifties and early sixties, not out of the portfolio. On the default order the premium in 2026 is $29,906 of their $70,000. If they paid it and kept the rest of their spending the same, they'd draw more each year, and these runs don't test that.

So which plan?

For this couple, the marketplace years are a tax problem with a floor and a ceiling, and their income needs to stay between $21,150 and $84,600 (rising with inflation) every year until 2032.

• Don't follow a bracket rule through these years. Filling the 12% bracket puts them over the cliff in every year, and they give up the whole credit.
• Watch the years with too little income. The default order spends the brokerage account first, which is usually good for taxes and leaves this couple with no credit for three years. A modest conversion in those years earned them $25,236 of credit in the first year alone.
• Remember the five-year wait. Converted money comes back tax-free five years later, and spending it can push income under the floor again. The plan has to cover all seven years, not just the first.
• Know the headroom. On the default order, once the IRAs pay the bills, their income sits $15,723 below the cliff in 2029. A conversion in those years has to fit inside that gap together with the extra IRA withdrawals that pay its tax, which count as income too. Go over and the whole credit goes. We didn't run a conversion sized that way.

The rest of the plan is in good shape: the money lasts in both stress tests and in 99.6% of historical start dates. The ACA decision is worth tens of thousands of dollars of their spending money, and it's made one year at a time.

What these runs leave out

• Their real premium. We used a national average benchmark for a 40-year-old, age-rated on the federal default curve and grown with inflation. The benchmark in their county could be well above or below it, and some states use their own age curves. Look up the second-lowest-cost silver plan on HealthCare.gov for your ZIP code and age before you plan around a number.
• Income under the floor, and Medicaid. Below 100% of the poverty line Plan Mirror gives no credit and stops there. The IRS rules are softer than that in one case: if the marketplace estimated at enrollment that the household's income would be at least 100%, and advance payments of the credit were made, the household can generally keep the credit even when its actual income ends up lower. That protects an honest estimate that turns out too high, not a plan to live below the floor. Whether a household under the floor qualifies for Medicaid depends on the state, and we didn't model that either.
• The plan they'd actually buy. We assumed the benchmark silver plan. A cheaper plan costs less after the same credit; a gold plan costs more.
• Next year's rules. These runs use the 2026 coverage year's poverty line, premiums and contribution percentages. Later years grow the poverty line and premiums with inflation and keep the 2026 percentages. Coverage for 2027 has its own poverty line, percentages and premiums, which we didn't use.
• A plan built around the window. We tested two simple conversion plans. Plan Mirror can't yet target an income level each year, so a third approach, converting just enough each year to stay inside the window, is untested here. We don't know from these runs how it would compare.

What we'll test next. Conversions sized to the ACA window year by year, once Plan Mirror can target an income. And the same couple with their real county premium, rather than a national average.

Every figure on this page comes from 3 recorded cases run together in Plan Mirror under 2026 tax rules. Each was run through the plan with its two stress tests, a historical replay of U.S. stock returns and inflation from 1926 Q3 to 2026 Q2, and 1,000 Monte Carlo paths. It is an illustration, not financial, tax, investment or legal advice. See the methodology.

Related scenarios

• 72(t) vs the rule of 55: one couple at 55, three ways into the 401(k)
• Can I retire at 55 with $2 million?
• Can I retire at 60 with $1 million?