The ACA Cliff Costs an Early-Retired Couple an Average of $215,469. Sequencing Withdrawals Avoids $207,055 of It.

QuantCalc Research · Published April 11, 2026 · Updated October 12, 2026 · Engine aca-cliff-engine-3.0.0 · 80,000 simulated paths

Above 400% of the federal poverty level there is no limit on repaying advance premium tax credits, so one dollar of modified adjusted gross income over the line in 2026 can cost the entire year’s credit. We ran 80,000 Monte Carlo paths on the QuantCalc planner’s own engines to measure the damage — and how much of it the order of withdrawals avoids.

What is the ACA subsidy cliff for early retirees in 2026?

Key numbers — each row links to the section that carries the full table. Updated 2026-10-12: rebuilt on the planner’s engines (what changed).

Why we ran this

For the 2026 coverage year the premium tax credit ends at 400% of the federal poverty level: the applicable-percentage table in IRS Rev. Proc. 2025-25 stops there, and the limit on repaying advance credits in 26 U.S.C. 36B(f)(2)(B) applies only to households below it. A household whose income ends the year above the line repays every dollar of advance credit it received.

How big is that dollar? On the engine’s national-average benchmark premium, the credit a household gets exactly at the line grows with age, because the benchmark is age-rated and the required contribution is fixed:

A household that crosses the line every year from 55 to 64 repays the sum: $88,090 for a single person, $216,620 for a couple.

What we modeled

Each profile is a 55-year-old household beginning a 10-year bridge to Medicare at 65. All four use the same portfolio allocation, the same account mix (60% Traditional, 15% Roth, 25% taxable at 50% cost basis) and the same return paths; they differ only in household size, starting portfolio and annual spending.

• Naive behavior — spend from the Traditional IRA/401(k) first until empty, then Roth, then taxable.
• Tax-optimized behavior — spend from the taxable account (half of each withdrawal is a long-term gain) and Roth first, and leave the Traditional account for after Medicare.

Profile-by-profile results

“Cliff yrs 1–3” is the share of paths whose MAGI crosses 400% FPL in any of the first three bridge years; “Ever crossed” in any of the ten. “Mean repayment” is the mean 10-year total of credit repaid. “Value of planning” is the difference between the two behaviors’ mean repayment. All in 2026 dollars.

Reading the numbers

• Naive repayment sits at the full credit. A naive household crosses every year, so its repayment is close to the $88,090 (single) or $216,620 (couple) sum above; a path falls short of it only in years when the Traditional account can no longer cover the spending.
• Optimized “ever crossed” rates are not zero. 19.4%–42.2% of optimized paths touch the cliff in at least one later year, as the taxable account runs down and spending shifts to the Traditional account. The mean cost stays small because none of those crossings falls in the first three years.
• Value of planning scales with household size. Singles save $78,311–$84,390; couples $196,295–$207,055, because a couple’s benchmark covers two people while the line rises only from $62,600 to $84,600.

Methodology

Every number comes from the QuantCalc planner’s engines; the study code only moves money between three accounts in the order each behavior prescribes. The full output is published as open data under a CC0 public-domain dedication: results.json.

Withdrawal-order logic is simplified on purpose: “naive” and “optimized” are fixed orders, not the year-by-year plan a real household would choose. Partial Roth conversions, state tax and IRMAA are not modeled. The planner’s ACA bridge optimizer builds that year-by-year plan for a specific household.

Changelog

• 2026-10-12 — rebuilt on the planner’s engines (engine aca-cliff-engine-3.0.0). The previous edition (aca-cliff-mc-2.0.0) used its own return generator, its own tax brackets, a flat national benchmark at every age ($14,400 a year single, $28,800 couple) and raised spending and the poverty line 2.5% a year against inflation-adjusted returns. Value of planning, previous edition → now:

  Profile	Previous	Now
  A	$77,444	$78,311
  B	$85,833	$84,390
  C	$213,290	$207,055
  D	$197,662	$196,295

• 2026-04-11 — first edition.

What this means for planners and retirees

If you are within five years of early retirement and have not modeled MAGI against the 400% FPL line for each bridge year, this is the number to run:

1. Map your expected MAGI in year 1 of retirement: Traditional withdrawals, Roth conversions, capital gains, interest, dividends, business income.
2. Know your 400% FPL number for your household size: for 2026 coverage, $62,600 (single) or $84,600 (couple).
3. Source as much spending as possible from taxable-account gains (where the cost basis dampens MAGI) and tax-free Roth distributions.
4. Size Roth conversions to stay under the line, and remember that the IRMAA threshold at 65 becomes the next cliff.
5. Test the plan across return sequences, not a single deterministic spreadsheet.

The ACA Cliff Calculator shows whether a given MAGI crosses the 2026 threshold and what it would cost; the Stress Test runs Monte Carlo paths on your own portfolio and withdrawal plan.

Frequently asked questions