Modeling IRMAA, the ACA Cliff, and Roth Conversions in Client Plans

Most retirement projections treat taxes as a flat drag — a single effective rate applied to withdrawals. For a high-level estimate that is fine. For a client in the years between early retirement and required minimum distributions, it misses the most consequential planning decisions of the decade, because those decisions are driven not by average tax rates but by thresholds: the income lines that, when crossed by a single dollar, trigger a step change in cost.

This guide explains why advisor-grade retirement plans need to model three threshold effects — IRMAA, the ACA subsidy structure, and Roth conversion headroom — and how they interact in a single client plan.

Why thresholds break flat-tax models

Run your own numbers — FREE

10,000 Monte Carlo simulations. Forward-looking forecasts from BlackRock, JPMorgan, Vanguard, GMO, Schwab, Invesco. No account needed.

Try QuantCalc Free →

A flat effective-rate model assumes the marginal cost of one more dollar of income is constant. Around a threshold it is not. Cross the wrong line by a dollar and the client can owe hundreds or thousands more — not on that dollar, but as a discrete penalty. Three of these matter most in retirement:

• IRMAA (Income-Related Monthly Adjustment Amount). Medicare Part B and D premiums step up at defined modified-AGI brackets. The surcharge is a cliff, not a ramp: one dollar over a bracket boundary applies the full higher premium for the year. And because IRMAA uses a two-year lookback, the income that triggers it was earned before the client felt the consequence.
• The ACA premium structure. For clients retiring before 65, marketplace premium assistance scales with income, and the structure around the upper income limit can make an extra dollar of realized income surprisingly expensive in lost assistance. A Roth conversion that looks free on a flat-tax model can quietly raise the client's net health-insurance cost.
• Roth conversion headroom. The flip side: in low-income years (often early retirement, before Social Security and RMDs), there is room to convert traditional balances to Roth at low marginal rates — but only up to the next threshold. The planning question is "how much can we convert before we hit the next cliff," which a flat model cannot answer because it does not know where the cliffs are.

How they interact

These do not operate independently. A Roth conversion raises modified AGI, which can simultaneously push the client toward an ACA assistance reduction (before 65) and, two years later, into a higher IRMAA bracket (after 65). The same dollar of conversion income can be taxed once and penalized twice. A plan that models conversions without modeling their downstream effect on IRMAA and ACA will recommend conversions that are not actually optimal.

The corollary is that the sequence matters. Converting aggressively at 62 may be worth a small ACA cost if it drains the traditional balance enough to keep RMDs — and the IRMAA they trigger — lower at 73. Whether that trade pays off depends on the client's specific balances, the conversion amounts, and the thresholds in force, which is exactly the kind of question a simulation can answer and a rule of thumb cannot.

What this means for the plan

For advisors, the takeaway is not that every plan needs to chase the last dollar of conversion efficiency. It is that a plan presented as tax-aware should actually model the thresholds it implicitly trades against. A probability-of-success number computed on a flat tax rate is answering a slightly different question than the client thinks it is — it is silent on the threshold costs that often dominate the early-retirement decade.

QuantCalc models these income-threshold effects inside the Monte Carlo engine rather than bolting them on afterward, so a conversion's effect on IRMAA brackets and ACA assistance shows up in the simulated outcomes, not just in a separate side calculation. Because the modeling lives in the engine, the threshold costs are reflected across all simulated paths — including the bad-market scenarios where a mistimed conversion hurts most.

Putting it in front of the client

When threshold modeling drives a recommendation, the client report should make the logic visible: which conversions, in which years, and what the plan assumed about the brackets. This is also a documentation point — the planning file should record the threshold assumptions alongside the return and inflation assumptions, since they materially shaped the advice. (See the companion post on what to document in the fiduciary plan file.)

QuantCalc's Advisor PRO tier ($249/year) carries these assumptions into the white-label report and the methodology supplement, so the threshold logic behind a Roth conversion recommendation is documented rather than implicit. You can see the format in the live sample at quantcalc.app/advisors.

Frequently asked questions

Do I really need threshold modeling, or is a flat tax rate close enough?

For a rough estimate, a flat rate is fine. For the early-retirement decade — when Roth conversion, ACA, and IRMAA decisions cluster — a flat rate hides the exact trade-offs the client is paying you to manage. The thresholds, not the average rate, drive those decisions.

Why does the two-year IRMAA lookback matter for planning?

Because the income that triggers a surcharge is earned two years before the higher premium is charged. A plan has to look ahead: a conversion at 63 can raise Medicare premiums at 65. Modeling the lookback explicitly is what lets the plan time conversions to avoid an avoidable bracket.

Does QuantCalc model these inside the simulation?

Yes. IRMAA brackets, ACA assistance effects, and Roth conversion headroom are modeled within the Monte Carlo engine, so their costs appear across all simulated paths rather than as a separate, deterministic side estimate.

QuantCalc is an independent educational tool, not affiliated with, endorsed by, or sponsored by any referenced firm. Not financial advice. Tax thresholds change; verify current figures before relying on them.