The Roth Conversion Ladder under SECURE 2.0 — modeling the post-2033 RMD-age regime

The SECURE Act 2.0 of 2022 raised the Required Minimum Distribution start age from 72 to 73 effective 2023, with a further increase to 75 scheduled for 2033. The change has structural consequences for the Roth conversion ladder strategy that retirement-planning platforms have spent the last two decades calibrating around.

For a household retiring at 65, the pre-SECURE 2.0 conversion window was 65 to 72 — seven years. The post-2033 window is 65 to 75 — ten years. That's a 43% expansion of the window during which conversions can happen at favorably-low rates before RMDs force ordinary-income recognition.

The expanded window doesn't just mean "more years to convert." It means a fundamentally different optimization problem because the IRMAA tier interactions, the Social Security taxation thresholds, and the surviving-spouse trajectory all unfold across this longer window. Retirement-planning platforms that ship pre-SECURE 2.0 logic — claiming "we handle Roth conversions" — are systematically advising the wrong ladder for households retiring after 2033.

This article is the working note on what the data model and projection logic need to support to advise the post-SECURE 2.0 regime correctly.

The mechanics, refreshed

A Roth conversion is a distribution from a traditional IRA or 401(k) immediately rolled to a Roth IRA. The distributed amount is taxable at the current ordinary-income rate. After conversion, the converted balance grows tax-free in the Roth and is not subject to RMDs.

The strategy: in the gap years between retirement (when ordinary income drops) and RMD age (when forced distributions begin), convert traditional balances to Roth at a lower marginal rate than the projected RMD-period rate. The lifetime tax savings is the rate-arbitrage: today's lower rate vs. RMD-period higher rate, applied to the converted balance, compounded by the tax-free Roth growth.

Three rate environments shape the calculation:

1. Current marginal rate during the gap year — typically lower than peak earning years because earned income has dropped.
2. Projected RMD-period marginal rate — driven by the household's projected RMD amount, projected Social Security receipts, and other income sources. Often higher than the gap-year rate.
3. Surviving-spouse marginal rate — typically materially higher than joint rate due to single-filer brackets and reduced Social Security. The surviving-spouse trajectory often pushes RMD-period rates higher than households project.

The optimal conversion amount each gap year is the amount that fills the bracket up to a defined ceiling — typically the top of the 22% or 24% bracket in current rules, or up to an IRMAA tier ceiling, whichever bites first.

The IRMAA tier interaction (still the hardest seam)

The Income-Related Monthly Adjustment Amount (IRMAA) for Medicare Part B and Part D premiums is computed on the household's MAGI from two years prior. Crossing an IRMAA tier means higher Medicare premiums for both spouses for one year, with the surcharge applying twelve months from the time the premium tier is set.

The 2026 IRMAA tier thresholds (updated annually for inflation) are roughly:

• $206,000 MAGI for married filing jointly: standard premium
• $258,000: tier 1 surcharge (~$70/month additional per spouse)
• $322,000: tier 2 surcharge (~$175/month additional per spouse)
• $386,000: tier 3 surcharge (~$280/month additional per spouse)
• $750,000: tier 4 surcharge (~$385/month additional per spouse)

For a married couple in 2026 crossing from tier 0 to tier 1, the IRMAA cost is roughly $1,680/year combined (both spouses), starting in 2028 (the two-year lookback).

The optimization complication: a Roth conversion that increases this year's MAGI affects the household's IRMAA tier two years out. A platform that optimizes conversions on this-year tax math without the IRMAA propagation will systematically over-convert in years where IRMAA tier crossings produce ongoing cost.

function projectConversionWithIRMAA(
  household: Household,
  conversionAmount: number,
  conversionYear: number
): ConversionProjection {
  const baseMAGI = projectMAGI(household, conversionYear);
  const newMAGI = baseMAGI + conversionAmount;
  
  const baseTier = computeIRMAATier(baseMAGI);
  const newTier = computeIRMAATier(newMAGI);
  
  const tierCrossing = newTier > baseTier;
  const irmaaImpactYear = conversionYear + 2;
  const irmaaImpactAmount = tierCrossing 
    ? annualIRMAACost(newTier) - annualIRMAACost(baseTier)
    : 0;
  
  // Conversion benefit
  const taxOnConversion = conversionAmount * marginalRate(household, conversionYear);
  const projectedTaxAvoided = projectFutureTaxAvoidance(
    conversionAmount, household, horizonYears
  );
  
  return {
    net_lifetime_benefit: 
      projectedTaxAvoided - taxOnConversion - irmaaImpactAmount,
    irmaa_tier_crossed: tierCrossing,
    recommended_amount_below_tier: tierCrossing 
      ? findAmountBelowTier(baseTier, baseMAGI, conversionAmount) 
      : conversionAmount
  };
}

The platform should identify, for each conversion year, the IRMAA tier ceiling and surface the recommended conversion that fills the bracket up to but not over the tier. Crossing the tier should require explicit user approval after seeing the cost.

The Social Security taxation seam

A second complicating seam: Social Security taxation depends on provisional income, which itself depends on Social Security receipts.

Provisional Income = AGI + Tax-Exempt Interest + 50% × Social Security Benefits

If provisional income exceeds $32,000 (joint), up to 50% of Social Security is taxable. Above $44,000 (joint), up to 85% is taxable.

For a household receiving Social Security, a Roth conversion that increases AGI propagates into the Social Security taxation calculation. The taxable share of SS goes up; the household's effective marginal rate on the conversion is higher than the headline bracket rate.

A common error: optimizing to "fill the 22% bracket" when the household is in the 50%-to-85% SS-taxation transition. The effective marginal rate can be 30-35% in this transition, not 22%. Conversions that look attractive at 22% are unprofitable at 35%.

The platform should compute the effective marginal rate at each MAGI level — including the SS phase-in — and use that rate for conversion decision-making.

The surviving-spouse trajectory

Married households face a structural rate transition at the first spouse's death. Single-filer brackets compress; one Social Security benefit is lost; the surviving spouse may have higher RMDs (combined account balances now subject to single-life RMD calculation).

For a couple expecting modest joint-life income but a high surviving-spouse income, conversions during the joint-life window are particularly attractive — converting at joint-filer rates today saves taxes at the higher single-filer rates the surviving spouse will face.

The platform must explicitly model the surviving-spouse trajectory. Many do not, treating the household as joint-filing forever. The error systematically under-recommends conversions for couples whose surviving-spouse trajectory implies higher future rates than today's joint rate.

SECURE 2.0's specific changes that change the optimization

Beyond the RMD age increase, SECURE 2.0 added several changes that affect Roth strategy:

• Roth-only catch-up for high earners (≥$145K wages, indexed): catch-up contributions to 401(k)s for these earners must go to the Roth bucket, not pre-tax. This shifts pre-retirement Roth balances upward, reducing the gap-year conversion need.
• 529-to-Roth rollover: up to $35K lifetime can roll from 529 to Roth IRA for the beneficiary, subject to 15-year-old account requirement and annual limit. Adds Roth bucket capacity for younger beneficiaries.
• Spousal RMD election: the surviving spouse can elect to use the deceased spouse's RMD age. Useful when the surviving spouse is younger.
• Designated Roth account RMD elimination (effective 2024): Roth 401(k) accounts no longer subject to RMDs during the participant's lifetime. Aligns with Roth IRA treatment.

Each of these changes may shift the optimal conversion amount, timing, or vehicle. A retirement-planning platform must reflect them in its current calculations.

The optimal ladder structure

The optimal conversion ladder for a typical post-SECURE 2.0 retiree (retirement at 65, RMDs starting at 75 post-2033):

• Years 65-66: large conversions to fill bracket up to IRMAA tier ceiling. No SS yet (delayed claiming), so MAGI is lowest.
• Years 67-69: SS may have started for one spouse; manage SS-taxation threshold. Conversions sized to fill bracket up to the lower of bracket ceiling or IRMAA tier.
• Years 70-74: both SS streams active; conversions tighter due to higher base income. Strategy may shift to focus on filling lower brackets only.
• Year 75 onward: RMDs start. Conversions become less attractive (incremental conversion stacks on top of RMD income, often pushing into higher brackets and tiers). Strategy typically shifts to "harvest losses, don't convert."

The platform should produce a year-by-year conversion plan with the projected impact on lifetime tax burden, the IRMAA tier risk per year, and the Social Security taxation thresholds.

What the test corpus needs

A test corpus for Roth conversion modeling under SECURE 2.0 needs:

• Pre-retirees in their late 50s and early 60s (planning case)
• Recent retirees in the gap years (active conversion case)
• Retirees just before RMD age (last-conversion-window case)
• Retirees with mixed traditional / Roth balances (already-laddered case)
• Households with substantial Social Security (SS-taxation seam)
• Households across IRMAA tier thresholds (tier-management case)
• Couples with high spousal income disparity (surviving-spouse trajectory)
• Multi-state retirement scenarios (CA → FL conversion timing)
• Households with delayed SS claiming (lower MAGI in early gap years)

Each scenario produces a structurally different right answer for the conversion ladder.

The synthetic-corpus value

Realistic retirement-data corpora must include the longitudinal projection — projected Social Security amounts (with claiming-age sensitivity), projected RMD trajectory under SECURE 2.0, projected IRMAA tier per year, projected surviving-spouse transition. Without these projections, the conversion-optimization engine has no inputs to optimize against.

The synthetic corpus serves as the validation infrastructure for the platform's projection engine itself. By comparing platform outputs to corpus expected values across scenarios, regression-testing the platform against the corpus catches the bugs that ship invisibly between engine versions.

Roth conversion modeling is one of the highest-leverage retirement-planning surfaces and one where SECURE 2.0 has materially shifted the optimization. Platforms that haven't updated their logic are systematically advising the wrong ladder for the most consequential decade of their retirees' tax lives. The data model and projection logic to handle the new regime are demanding but enumerable; the platforms that get it right differentiate sharply in the retirement-income segment.