Roth vs Traditional Catch-Up in 2026: When SECURE 2.0 §603 Actually Hurts You — Decision Framework, Break-Even Math, and the 5-8% Edge Cases

January 1, 2026 activated SECURE 2.0 Act §603 for real. After two years of administrative transition relief under IRS Notice 2023-62,^[1] any catch-up contribution made by a participant whose prior-year Social Security wages from the plan sponsor exceeded $150,000 (indexed from the statutory $145,000 baseline) is now required to be a designated Roth contribution rather than a pretax elective deferral.^[2] The base §414(v)(2)(B)(i) catch-up limit for 2026 is $8,000. The §414(v)(2)(E)(i) age-60-through-63 super catch-up, added by SECURE 2.0 §109, is $11,250.^[3] A high-earning 61-year-old in 2026 who wants to max their catch-up is contributing $11,250 that used to be tax-deductible into an after-tax Roth bucket by statutory command, not by choice.

The industry consensus around §603 has settled into a comfortable talking point: Roth catch-up is fine, and for most high earners it's actually better because it locks in today's tax rate, avoids Required Minimum Distributions on the Roth 401(k) balance under SECURE 2.0 §325, escapes the Medicare IRMAA cliff structure at retirement, and produces tax-free intergenerational transfers under the 2020 SECURE Act ten-year non-spouse beneficiary rule.^[4] That consensus is directionally correct — for roughly 92 to 95 percent of §603-covered participants. But for a specific 5 to 8 percent of cases, §603 forces the mathematically inferior choice. Vanguard's 2025 How America Saves report shows about 26 percent of §414(v)-eligible participants at the plan-sponsor tier where §603 will apply — roughly 3.8 to 4.5 million U.S. workers.^[5] Even a 5 percent misalignment rate is 190,000 to 225,000 participants who are being pushed into the wrong tax bucket by an eight-figure-annual dollar amount.

This guide is the decision framework for figuring out which side of the break-even you sit on. It walks the §603 mandatory-Roth rule, the exact break-even math distilled to one formula, the five variables that determine the answer (current marginal rate, retirement marginal rate, state-tax delta, filing-status transitions, IRMAA exposure), the two shortcut cases that answer instantly, the specific fact pattern where §603 actively hurts, four workarounds when §603 forces the wrong side of the break-even, three worked case studies at $180K / $260K / $420K wages, six framework mistakes to avoid, and an 8-item checklist. Model the arithmetic as you read using the Roth IRA calculator, the 401(k) calculator, and the income tax calculator, which handle the marginal-rate layering and after-tax equivalence math directly.

1. What §603 actually does — and who it captures

SECURE 2.0 Act §603 amended §414(v) to add a new subsection (7) that reclassifies catch-up contributions for high earners. The statutory text reads, in relevant part: "In the case of an applicable employer plan, no additional elective deferrals may be made pursuant to this subsection by a participant whose wages (within the meaning of section 3121(a)) for the preceding calendar year from the employer sponsoring the plan exceeded $145,000 (as adjusted under paragraph (3)(B)) unless such additional elective deferrals are designated Roth contributions."^[2] The reference to §3121(a) is the Social Security wage base definition — Box 3 of the Form W-2. The $145,000 threshold is indexed under §414(v)(2)(C) using the cost-of-living methodology; the IRS confirmed the 2026 threshold of $150,000 in Notice 2025-67 released September 2025.^[6]

Four features of the rule matter for practical planning. First, the wage test is prior-year, not current-year. A participant's 2026 §603 status is determined by 2025 Form W-2 Box 3 wages from the plan sponsor. A participant who joined a new employer in November 2025 with $180,000 of 2026 salary but no 2025 W-2 wages from that employer is exempt from §603 for 2026 catch-up contributions from that plan — the "preceding calendar year" wage from the plan sponsor is zero. Second, the test is plan-sponsor-specific. A participant with $130,000 from Employer A (the plan sponsor) plus $50,000 of self-employment income does not trigger §603 for Employer A's plan because self-employment income is not Employer A wages.^[7] Third, the test uses the Social Security wage base cap by default — Box 3 caps at $176,100 for 2025 wages reported on the 2025 W-2 used for 2026 testing — though plan sponsors may elect the uncapped Medicare wages (Box 5) method under Notice 2023-62 clarifications. Fourth, the rule applies to §401(k), §403(b), and governmental §457(b) plans; it does not apply to SEP-IRA, SIMPLE-IRA, or governmental §457(b) plans of state and local subdivisions that treat catch-up contributions outside the §414(v) framework.

The scope of "any catch-up contribution" is broader than most participants realize. The plain-language reading of §603(a) captures both the base §414(v)(2)(B)(i) catch-up ($8,000 for 2026) and the §414(v)(2)(E)(i) super catch-up for participants aged 60, 61, 62, or 63 ($11,250 for 2026, calculated as the greater of $10,000 or 150 percent of the base catch-up).^[3] A §603-covered 61-year-old participant faces mandatory Roth treatment on the full $11,250 super catch-up, not just the base $8,000 — which is the specific window when many participants are trying to close pre-retirement funding gaps at the highest marginal rates of their careers.

2. The break-even math — one formula

The core Roth-vs-pretax comparison collapses to a single equation once you strip away the noise. A dollar of pretax contribution today saves you the current marginal tax rate (call it MTR_now) in current-year taxes, grows tax-deferred, and is taxed at the retirement marginal rate (call it MTR_ret) on withdrawal. A dollar of Roth contribution today pays MTR_now in current-year taxes, grows tax-free, and is withdrawn tax-free. Comparing the two on an equivalent after-tax basis — meaning you contribute the pretax equivalent of the same take-home cash flow — Roth wins whenever MTR_ret exceeds MTR_now, pretax wins whenever MTR_now exceeds MTR_ret, and the two tie whenever MTR_now equals MTR_ret.

Two subtleties that most simplified explanations skip. First, MTR is total marginal, not just federal. State income tax at withdrawal is imposed on Traditional distributions and not on Roth distributions, so a state-tax delta between working state and retirement state is part of MTR_now vs MTR_ret. Second, marginal rate at retirement is not a single number — it's the marginal rate on the specific Traditional distribution being modeled. Layering a $50,000 Traditional distribution on top of $30,000 of Social Security and $15,000 of taxable dividends may exit at 12 or 22 percent federal depending on total income; layering another $50,000 in the same year may exit at 24 percent because the first tranche filled the 22 percent band.^[8]

The intuition most consumer-facing content offers — "Roth if you'll be in a higher bracket later, pretax if lower" — is directionally correct but obscures the state-tax and IRMAA layers that swing the answer for real participants. A California resident at 32 percent federal + 10.3 percent state = 42.3 percent effective MTR_now who relocates to Florida at retirement and draws distributions at 24 percent federal + 0 percent state = 24 percent MTR_ret sees a 18.3-percent break-even margin in favor of pretax. That's the fact pattern §603 blocks: the participant is high-earning today, will retire in a lower-tax state, and would have exited Traditional distributions in a lower federal-plus-state combined bracket. The Roth path locks in the higher combined bracket forever.

3. The five decision variables — and which one dominates for you

Every real Roth-vs-pretax calculation is a projection over four decades, and the accuracy of the answer depends on how many of the five key variables you can pin down credibly. In order of typical impact for a §603-covered participant:

Two of the five variables — current MTR and non-spouse heir plans — are known with high confidence at the decision point. Current MTR is a lookup: federal from IR-2025-176 for 2026, state from your state department of revenue, plus 3.8 percent Net Investment Income Tax on relevant components, plus 0.9 percent Additional Medicare Tax on wages above $200,000 (single) or $250,000 (MFJ).^[9] Non-spouse heir plans are typically clear from existing estate documents — if the participant's primary beneficiary designation is a spouse who is close in age, non-spouse legacy is a small factor; if the primary designation is adult children who will inherit at their own peak-earning ages, non-spouse legacy is a large factor.^[10]

The three medium-confidence variables — retirement MTR, state-tax delta, and IRMAA exposure — are the ones where projection error dominates the framework's honesty. A participant who says "I'll be in the 12-percent bracket at retirement" while planning to draw $180,000 per year from a mixed portfolio is projecting wrong; a participant who says "I'll move to Florida at 62" while owning a home in California with school-age grandchildren nearby is projecting wrong. The framework's value is not resolving these uncertainties — it's naming them so the participant makes an explicit projection and can update it later.

4. When §603 actively hurts — the specific fact pattern

The four-variable alignment that produces the "Roth is worse" outcome is precise enough to describe explicitly. All four must be true for the pretax-would-have-been-better verdict to hold:

Condition A: Current marginal rate is 32 or 35 percent federal. The 2026 federal 32 percent bracket runs from $206,700 to $394,600 for MFJ and from $206,700 to $256,225 for single filers. The 35 percent bracket runs from $394,600 to $751,600 MFJ and from $256,225 to $626,350 single.^[11] Add 3.8 percent Net Investment Income Tax on the participant's investment income component and 0.9 percent Additional Medicare Tax on wages above $200,000 (single) / $250,000 (MFJ) and the effective marginal on the catch-up dollar reaches 36.7 percent to 39.7 percent federal alone. This is the peak-earning band where pretax deductibility is worth the most in absolute dollars.

Condition B: Retirement plan involves a high-tax-to-low-tax state relocation. Nine states impose no state income tax on retirement distributions: Alaska, Florida, Nevada, New Hampshire, South Dakota, Tennessee, Texas, Washington, and Wyoming. Eight additional states have full or substantial retirement-income exemptions: Illinois, Iowa, Kentucky, Mississippi, Pennsylvania, and — with age-based limits — Georgia, South Carolina, and Utah.^[12] A participant currently residing in California (12.3 percent top marginal), New York (10.9 percent), New Jersey (10.75 percent), Hawaii (11 percent), Oregon (9.9 percent), or Minnesota (9.85 percent) who plans to relocate to a no-income-tax state at retirement is capturing 9.9 percent to 12.3 percent of state-tax savings on every pretax dollar that will be withdrawn post-move. Roth locks in that state tax today; pretax defers and dodges it. For a $11,250 super catch-up contribution, the state-tax delta alone is worth $1,110 to $1,380 per year permanently — before compounding.

Condition C: Projected retirement marginal rate lower than current. A participant who will retire without wage income, drawing from a mix of Social Security (taxed at 0-85 percent depending on provisional income), pension (fully taxable), Traditional IRA distributions (fully taxable), Roth IRA distributions (tax-free), qualified dividends (0/15/20 percent capital gains rates), and long-term capital gains (same) typically has a materially lower effective marginal on the specific Traditional 401(k) distribution than during peak earning. The 2026 single-filer 22 percent bracket runs from $47,150 to $100,525 and the 24 percent bracket from $100,525 to $191,950.^[11] A participant drawing $100,000 per year from Traditional balances on top of $30,000 of Social Security exits at 22 percent federal. Compare that to their 32 or 35 percent current bracket — the delta is 8 to 13 percentage points on top of the state delta.

Condition D: Small or no non-spouse legacy component. The most powerful argument for Roth at high income levels is the intergenerational tax-free transfer under SECURE Act §401's 10-year non-spouse beneficiary rule.^[13] A participant whose beneficiaries are all spousal (with the surviving spouse simply continuing to hold the Traditional balance as their own IRA) captures little of this benefit; a participant whose beneficiaries are adult children in their peak-earning years captures a lot. Condition D is the "small legacy component" side — beneficiary is spouse, or heirs are already low-earning, or the participant plans to consume rather than transfer the retirement balance.

When A + B + C + D are all true, mandatory Roth catch-up under §603 is the mathematically inferior choice by 3 to 6 percent of the catch-up dollar amount over the participant's lifetime. For a $11,250 super catch-up contribution, that's $337 to $675 per contribution year in NPV terms — modest per year but compounding over a 5-to-10-year peak-earning window into $2,000 to $8,000 of lifetime tax loss per participant.

5. Four workarounds when §603 forces the wrong side

If the framework identifies you as one of the misaligned 5-8 percent, four workarounds recover some or all of the pretax advantage that §603 blocks. None fully replaces the pretax catch-up — that option is legally foreclosed — but each captures a partial substitute.

Workaround 1: Load the base §402(g)(1) deferral pretax and take the §603 hit only on the catch-up

The §603 mandatory Roth rule applies only to catch-up contributions under §414(v) — not to the base §402(g)(1) elective deferral limit ($24,500 for 2026). A §603-covered participant retains full pretax election on the base $24,500 and only faces mandatory Roth on the $8,000 base catch-up or $11,250 super catch-up.^[14] The take-away: max the base pretax first ($24,500 saves $7,350 to $10,913 in federal tax at 30 to 44.5 percent effective MTR), then take the §603 Roth hit only on the smaller catch-up piece. This limits the affected dollar amount to $8,000 or $11,250 rather than $32,500 or $35,750. A participant who thinks §603 is bad for them and stops contributing base deferrals in protest loses the far larger pretax opportunity.

Workaround 2: After-tax voluntary contribution plus in-plan Roth conversion (mega backdoor Roth)

Plans that offer after-tax voluntary contributions plus in-service in-plan Roth conversion enable the mega backdoor Roth mechanic covered in the mega backdoor Roth 2026 guide. The §415(c) overall annual additions limit is $72,000 for 2026 (up from $70,000 for 2025).^[15] A participant contributing $24,500 in base pretax + $11,250 in Roth catch-up + up to $36,250 in after-tax voluntary contributions ($72,000 − $24,500 − $11,250 = $36,250, minus any employer match that occupies the §415(c) headroom) can convert the after-tax voluntary contributions to Roth 401(k) in-service. The tax posture is different from pretax catch-up — after-tax voluntary contributions are made with after-tax dollars — but the mechanic captures the pretax-equivalent growth on a dollar amount much larger than the catch-up limit.

Workaround 3: §409A Nonqualified Deferred Compensation election

Employers offering a §409A nonqualified deferred compensation (NQDC) plan let executives and high earners defer additional compensation beyond §401(k) limits. NQDC deferrals are pretax at the federal level (subject to FICA at the earlier of vesting or payment), grow tax-deferred, and are taxed at the ordinary rate on distribution per the participant's original election.^[16] For a §603-covered participant who wants pretax deferral beyond what the §401(k) framework allows, NQDC is the primary vehicle. Two caveats: (a) NQDC balances are unsecured obligations of the employer subject to bankruptcy risk, and (b) the initial deferral election is generally irrevocable and payment triggers are constrained. For high-tenure executives at financially stable employers, NQDC captures the pretax benefit §603 blocks; for mid-career participants at financially uncertain employers, the credit risk may outweigh the tax benefit.

Workaround 4: Post-retirement Roth conversion ladder

The fourth workaround inverts the timing: contribute the §603-required Roth catch-up now, then Roth-convert existing Traditional IRA and Traditional 401(k) balances at low income during retirement. The Roth conversion ladder guide covers the mechanic. A participant with a $1.5M Traditional 401(k) balance and $500K Roth 401(k) balance who retires at 60, moves to Florida, and lives on taxable-account cash while converting $80,000 per year from Traditional to Roth at the 12 or 22 percent federal bracket effectively moves pretax dollars into Roth at a much lower rate than they would have exited by RMD age.^{[17] This does not undo the §603 damage on the current catch-up contribution, but it captures the pretax-conversion opportunity on the much larger Traditional balance the participant is already sitting on. The conversion window is age 60 to age 73 (the 2026-plus SECURE 2.0 RMD age) — thirteen years to move dollars strategically.}

Elena plans to retire in Illinois. Critically, Illinois exempts qualified plan distributions from state income tax under 35 ILCS 5/203(a)(2)(F).[12] Her projected 2028 retirement MTR: drawing $95,000 per year from Traditional balances plus $32,000 Social Security, her federal marginal exits at 22 percent (2028 MFJ 22 percent bracket, indexed). Illinois marginal: 0 percent on qualified plan distributions. Total MTR_ret: 22 percent.

Elena's MTR_now vs MTR_ret differential: 29.85 percent − 22 percent = 7.85 percentage points. Pretax catch-up of $11,250 would have been worth $11,250 × 7.85 percent = $883 in first-order tax delta versus Roth, before compounding. The differential is real but modest. Adding the IRMAA analysis: her projected 2030 MAGI (two-year lookback from 2028) sits around $145,000 MFJ, comfortably within the second IRMAA tier ($211,000-$266,000 MFJ for 2026, adjusting for 2030 indexing).^[18] Adding $11,250 of Traditional distributions per year at the margin doesn't cross into the third tier. The IRMAA effect is neutral for her.

Case 2 is the boundary case: §603 costs Elena approximately $883 per year in first-order tax terms but delivers approximately $400 per year in IRMAA optionality (protects the MAGI headroom for future flexibility) and approximately $600 per year in non-spouse legacy value (her adult children in California and Colorado will inherit the Roth balance tax-free under the 10-year rule). Net: §603 is roughly neutral for Elena, arguably marginally beneficial. She should not attempt workarounds — the framework's honest verdict is that the mandatory-Roth path is about the same as the pretax path would have been. She saves the analysis effort and accepts the §603 treatment as-is.

Case 3: Marcus, age 62, executive at a New York hedge fund, $600,000 base + $1.2M bonus, MFJ, New York, $4.8M Traditional 401(k) balance across three prior employers, plans to retire to Florida in 2027

Marcus's 2025 W-2 Box 3 wages from the plan sponsor were $176,100 (capped). §603 applies for 2026. His 2026 combined effective marginal on catch-up dollars is 35 percent federal + 6.85 percent New York State + 3.876 percent New York City + 3.8 percent NIIT + 0.9 percent Additional Medicare Tax = 50.4 percent effective on the marginal catch-up dollar. He's in the age-60-63 super catch-up window and can contribute $11,250.

Marcus's 2028 retirement MTR (post-Florida move): drawing $220,000 per year from Traditional balances plus $58,000 Social Security, his federal marginal on the specific Traditional distribution exits at 24 percent (2028 MFJ 24 percent bracket, indexed). Florida marginal: 0 percent. Total MTR_ret: 24 percent.

Marcus's MTR_now vs MTR_ret differential: 50.4 percent − 24 percent = 26.4 percentage points. This is the fact pattern §603 hurts most: peak Manhattan effective marginal now, retirement in Florida later, MFJ retirement income at 24 percent federal on a large Traditional distribution. Pretax catch-up of $11,250 would have been worth $11,250 × 26.4 percent = $2,970 in first-order tax delta versus Roth, before compounding.

Marcus's workaround stack: (a) max base pretax deferral $24,500 at 50.4 percent effective marginal = $12,348 in current-year tax savings — this is the largest single lever. (b) Execute the full mega backdoor Roth mechanic — his plan permits after-tax voluntary contributions plus in-service in-plan Roth conversion. He contributes $36,250 in after-tax voluntary contributions (up to the §415(c) headroom), converts to Roth quarterly, capturing $36,250 of after-tax dollars into a tax-free growth wrapper.^[15] (c) His employer offers a §409A NQDC plan — he defers an additional $200,000 of his 2026 bonus pretax to be paid out over 5 years post-separation. This captures $200,000 × 50.4 percent effective marginal = $100,800 in current-year tax deferral, at the cost of unsecured-obligation credit risk to the employer. (d) Post-2027 Florida relocation, Marcus executes an aggressive Roth conversion ladder from the $4.8M Traditional balance, converting $200,000 per year at Florida's 0-percent state rate + 24 percent federal rate = 24 percent effective conversion cost, versus the 50.4 percent effective RMD cost he would have faced starting at 73 in New York.

Net verdict for Marcus: §603 costs him approximately $2,970 in the current-year first-order delta on the catch-up, but the workaround stack recovers 90-plus percent of the pre-§603 optimal because the base pretax deferral is much larger than the catch-up and the post-relocation Roth ladder converts pretax dollars at a lower rate than RMDs would have imposed. Marcus is materially better off after §603 than a passive "pay the required Roth and do nothing else" response would produce. The framework earned its keep by pointing him at workarounds 1, 2, 3, and 4 simultaneously.

7. Six framework mistakes to avoid

Mistake 1: Assuming your current federal bracket is your current effective marginal. A 32 percent federal bracket participant with investment income above the NIIT threshold and wages above the Additional Medicare Tax threshold has a much higher effective marginal on the specific catch-up dollar. Add 3.8 percent NIIT if you have net investment income (interest, dividends, capital gains, rental) above $200,000 single or $250,000 MFJ, and 0.9 percent Additional Medicare Tax on wages above the same thresholds. The stacking can push effective federal marginal to 36.7 percent and combined effective marginal above 45 percent in high-tax states.^[9]

Mistake 2: Assuming your retirement marginal is your retirement effective rate. The Roth-vs-pretax comparison uses marginal, not average or effective. Blending Social Security (partially taxable), Traditional IRA distributions (fully taxable), qualified dividends (0/15/20 percent), and long-term gains (same) at retirement produces a low average rate. But the specific pretax distribution being modeled exits at whatever marginal bracket it fills. A participant with $150,000 of retirement income at 18 percent effective average may be at 24 percent marginal on the next dollar of Traditional distribution. The marginal number is what enters the framework.^[8]

Mistake 3: Ignoring the state-tax delta. State income tax on Traditional 401(k) distributions is imposed by the state of residence at time of distribution, not the state where the deferral was earned. A participant who defers pretax while in California pays no California tax on the deferral, and if they relocate to Florida before distribution, pays no Florida state tax on the distribution either. A participant who defers Roth while in California pays California tax on the wage income that funded the Roth deferral and receives no state-tax recovery on the tax-free Roth distribution later. The state-tax delta is one of the largest single levers and the most frequently omitted variable in consumer-facing comparisons.

Mistake 4: Underweighting IRMAA. The Medicare Part B and Part D IRMAA surcharge structure adds up to $5,326.80 per person per year at the highest 2026 income bracket. For a high-balance retiree drawing $200,000+ per year from Traditional balances plus Social Security plus taxable investment income, IRMAA is a meaningful additional cost of Traditional distributions that is not captured in the marginal-rate comparison. Roth distributions do not increase MAGI for IRMAA purposes.^[18]

Mistake 5: Forgetting about beneficiary structure. The 10-year non-spouse beneficiary drawdown rule under SECURE Act §401 dramatically changes the intergenerational-transfer value of Roth vs Traditional balances. An adult child inheriting a $500,000 Traditional 401(k) at their peak-earning age faces $500,000 of ordinary income spread over 10 years — potentially $150,000 to $185,000 of federal tax if the heir is in the 32-37 percent bracket. The same $500,000 Roth 401(k) transfers tax-free. Participants with non-spouse beneficiary designations should heavily weight this factor in favor of Roth even before the §603 discussion.^[13]

Mistake 6: Not maxing base pretax first. The single most valuable §603 workaround is maxing the $24,500 base §402(g)(1) deferral pretax. A participant who is skeptical of Roth catch-up and stops contributing entirely loses the far larger pretax opportunity on the base deferral. The correct posture is: max base pretax, accept required Roth on the smaller catch-up piece, and consider the four workarounds for the residual imbalance.

8. Your 8-item 2026 catch-up decision checklist

9. Frequently asked questions