AGI vs MAGI: Why Modified Adjusted Gross Income Has Multiple Definitions

Possibly. The answer depends on which MAGI he calculated. Let’s check.

For a married couple filing jointly, the 2026 Roth IRA contribution phase-out starts at approximately $236,000 (subject to 2026 IRS final figures). If your MFJ MAGI is $215,000, you’re below the phase-out start and entitled to full Roth contributions ($7,000 per spouse under 50, or $8,000 if 50+).

Wait — but the IRS uses the MFJ start at $236K and you’re at $215K. You’re below the threshold by $21K, so you’d get FULL contribution, not partial. Why did your CPA tell you you can only contribute partially?

Possibility 1: Your CPA used the NIIT MAGI by mistake. The NIIT threshold is $250K MFJ, but with different add-backs. Maybe he confused thresholds.

Possibility 2: Your CPA used the single filer threshold by mistake. Single phase-out is around $150K-$165K; if he thought you were single, $215K would put you above the upper limit, blocking Roth entirely.

Possibility 3: Your CPA calculated MAGI as $215K but the actual MAGI for Roth IRA purposes is higher (because he didn’t add back items that should be added back). The Roth MAGI = AGI + traditional IRA deduction + student loan interest deduction + foreign earned income exclusion + foreign housing exclusion + savings bond exclusion. If your AGI is $215K but you also had a $10K traditional IRA deduction, the Roth MAGI is $225K — still below $236K start. Or if you had a foreign earned income exclusion of $80K (typical for an expat year), Roth MAGI could be $295K — above the upper end.

Ask your CPA specifically: ‘For Roth IRA phase-out purposes, what did you calculate as my MAGI, and what add-backs did you include?’ Get him to walk through the math line by line.

If his MAGI calculation is correct at $215K MFJ, you should be able to contribute the full $7,000/$8,000 per spouse for the year. If his calculation includes correct add-backs that bring you above $236K, then you’re in phase-out territory and partial contribution applies; if above $246K, no direct Roth at all (but backdoor Roth path is open).

Let’s run the partial contribution math for your scenario as stated. If MAGI is $215K (below $236K start), full contribution = $7,000 per spouse = $14,000 total for the couple. If MAGI were say $241K (in the phase-out range), partial contribution = $7,000 × (1 − ($241,000 – $236,000) / $10,000) = $7,000 × 50% = $3,500 per spouse = $7,000 couple total.

Double-check by also considering: were any of these CPA messages about a future year vs. current year? Sometimes CPAs project forward and the projected income is higher than current. Or if you’ve already made a Roth contribution and then your year-end income spike pushed MAGI above threshold, your contribution is now excess and needs to be withdrawn (or recharacterized) by October 15 of the following year to avoid the 6% excess contribution penalty.

Final note: if direct Roth contribution becomes constrained (you’re in phase-out or above), the backdoor Roth path is still open. Make a non-deductible traditional IRA contribution of $7,000, then convert it to Roth. The conversion has no income limit. The contribution doesn’t generate a deduction, but you get Roth tax-free growth on the converted balance.

Get clarity from your CPA on which MAGI definition he applied and what specific add-backs (or not). Without that clarity, you’re guessing — and the consequences of wrong contribution amounts include the 6% excess contribution penalty if too much, or missing out on tax-advantaged Roth growth if too little.

No, because Roth IRA MAGI adds back the foreign earned income exclusion. Your MAGI for Roth IRA purposes is $250K, not $130K. You’re well into the MFJ phase-out (above the $246K upper bound) and cannot make a direct Roth contribution.

Here’s the math step by step:

Gross wages from London: $250,000 Foreign earned income exclusion under §911: $120,000 (assumed) AGI (line 11): $130,000 ($250K – $120K)

For Roth IRA contribution purposes, MAGI = AGI + foreign earned income exclusion + foreign housing exclusion + other add-backs.

Foreign earned income exclusion add-back: $120,000 Roth MAGI: $130,000 + $120,000 = $250,000

If you also claimed a foreign housing exclusion under §911 (which can add another $20K-$50K of exclusion for high-cost-of-living posts like London), add that too. Roth MAGI could be $270K-$300K for someone like you.

For single filer in 2026: phase-out is $150K-$165K. Your Roth MAGI of $250K is far above; no direct Roth allowed. For MFJ: phase-out is $236K-$246K. Your $250K is above the upper bound; no direct Roth.

Why does the IRS add back the foreign earned income exclusion? Policy reason: Congress designed the foreign earned income exclusion to give expats relief from US tax on foreign-source compensation, but didn’t want the exclusion to also qualify expats for other tax-advantaged accounts that have income limits. The add-back effectively says, ‘For purposes of determining whether you’re eligible for the Roth IRA contribution, treat your full income, not your post-exclusion income.’

Your options as an expat above the phase-out:

1. Backdoor Roth IRA. Make a non-deductible traditional IRA contribution of $7,000 (for 2026), then convert to Roth. Conversion has no income limit. The pro-rata rule applies — if you have pre-tax IRA balances (e.g., rollover IRA from a former 401(k)), the conversion is partially taxable. If your only traditional IRA is the new non-deductible contribution, the conversion is essentially tax-free except for any growth between contribution and conversion.

2. Mega Backdoor Roth (if available through your employer’s 401(k)). If your US employer’s 401(k) plan allows after-tax non-Roth contributions and in-plan Roth conversions, you can contribute up to the §415 limit (around $69K-$75K for 2026 depending on inflation adjustments) of pre-tax + employer + after-tax, with the after-tax portion convertible to Roth annually. This isn’t an IRA, it’s a 401(k) feature. Not all plans allow it.

3. Reduce your foreign earned income exclusion. Take the §911 exclusion as a more limited amount or claim foreign tax credit instead. If you have UK tax paid that exceeds your hypothetical US tax on the same income, the foreign tax credit may provide better tax efficiency than the exclusion AND lower your Roth MAGI. Run both calculations.

4. Wait until you return to the US. Once you’re no longer on foreign assignment and have no §911 exclusion, your Roth MAGI drops back to AGI. If your post-return income is below the phase-out, direct Roth contributions resume.

5. Backdoor Roth + UK pension. Some UK pension contributions (pension funds, ISA, etc.) are tax-advantaged in the UK but may have complex US tax treatment. Coordinate with both UK and US advisors.

One practical concern: many expats living abroad have trouble finding US-based Roth IRA custodians who will open accounts for non-US residents. Some custodians require a US address. If you’re in London long-term, this practical issue may be more limiting than the MAGI rule. Plan for it.

For your specific situation: backdoor Roth is the most straightforward workaround. Talk to your CPA about the exact mechanics and whether your existing IRA balances would trigger pro-rata complications. Our expat tax team handles these structures regularly for clients on London/Tokyo/Singapore assignments.

Great planning window. Your income dropped sharply with retirement, opening room for Roth conversions at lower tax rates. But you have to balance Roth conversion against ACA premium tax credit eligibility because they pull in opposite directions. Here’s the analysis:

Without any Roth conversion: AGI = $35K dividends + taxable Social Security. Assume Social Security is $30K with maybe $15K taxable (modest tax on lower-income retirees). Total AGI = ~$50K.

ACA PTC eligibility: MAGI for ACA = AGI + tax-exempt interest + non-taxable Social Security ($15K) + foreign exclusions = $50K + $15K = $65K MAGI for ACA. For a household of 2, 2026 FPL is around $21K; you’re at about 300% of FPL. PTC applicable percentage is around 6% of household income. You’d pay about 6% × $65K = $3,900 toward premiums; PTC covers the rest. Probably $8K-$15K of PTC per year depending on actual premium levels.

Now add a Roth conversion of $100K: AGI rises to $150K. ACA MAGI rises to $165K. At a household of 2 with $165K income, you’re around 750% of FPL. Under the Inflation Reduction Act extension (through 2025, status for 2026+ depends on Congress), there’s no income cliff — PTC continues but at 8.5% of household income. PTC would cover anything above $165K × 8.5% = $14,025 of premium. Likely still some PTC available, but smaller.

The trade-off: Roth conversion creates current-year tax cost at marginal rates (probably 22-24% federal bracket on the conversion) plus state tax. ACA PTC reduction is real money out of pocket for premiums.

Let’s optimize. Several considerations:

1. Marginal tax rate on Roth conversion. At $35K base AGI, you have substantial room to convert at low rates. The 12% federal bracket ends around $96K for MFJ in 2026 (projected). You could convert up to $96K – $35K = $61K at 12% federal rate. That’s a great rate. Beyond that, the rate jumps to 22% bracket.

2. Pre-Medicare ACA cost vs. post-Medicare cost. Medicare starts at 65. ACA PTC ends when you enroll in Medicare. So your ACA PTC concern is for years 60-64 (about 5 years). After 65, you’re on Medicare and the ACA isn’t relevant — but you have IRMAA, which is a different MAGI concern.

3. IRMAA 2-year lookback. Your 2026 tax return determines your 2028 Medicare premium. A Roth conversion in 2026 hits both your 2026 tax bill AND your 2028 Medicare premium. The Medicare IRMAA add-back is just AGI + tax-exempt interest, so a $100K Roth conversion in 2026 likely pushes 2028 IRMAA into a higher bracket.

4. RMDs at 73. If you don’t convert during the low-income window (60-72), traditional IRA balances grow and force larger Required Minimum Distributions starting at age 73 (under SECURE 2.0). RMDs at high balances can push you into higher brackets later, plus drive IRMAA up.

Optimization strategy:

Year by year through age 64: convert enough Roth to use the 12% bracket while keeping ACA MAGI manageable. Aim for $50K-$70K of conversion per year. Total of $250K-$350K over the 5 years before Medicare. This builds Roth balance significantly without sharp tax cost.

At age 65 (Medicare-eligible, ACA PTC ends): the ACA constraint releases, but IRMAA constraint binds (2-year lookback means your age-63 income affects age-65 premium). Continue moderate Roth conversions but consider IRMAA brackets.

Age 65-72 (Medicare years, pre-RMD): your concerns are (1) ordinary income tax on conversions, (2) NIIT if MAGI above $250K MFJ, (3) IRMAA tier crossings. Continue Roth conversions at amounts that respect these thresholds.

Age 73+: RMDs begin. Roth conversions get harder to do efficiently because the marginal rate is higher with RMDs already pushing you up. Most of the work should be done pre-73.

Specific math for your year 1 (current year):

– Current AGI: $50K. Add Roth conversion of $50K. New AGI: $100K. Still in 12% bracket (under $96K is 12%, $96K-$201K is 22%). Federal tax cost on $50K conversion: approximately $5K incremental. Plus state tax.

– ACA MAGI: $100K + non-taxable SS ($15K) = $115K. About 540% of FPL for household of 2. Under IRA extension through 2025, PTC applies at 8.5% of household income cap. Premium expectation is 8.5% × $115K = $9,775 contribution; whatever your unsubsidized premium exceeds $9,775 is your PTC.

– Compared to no conversion: your premium contribution might be 4-6% of $65K = $2,600-$3,900. So the conversion costs you about $6K-$7K of additional ACA premium contribution. Plus the $5K of federal tax. Total cost of $50K conversion: about $11K-$12K, or about 22-24% of the conversion amount.

– That’s roughly equivalent to the 22% federal bracket. The ACA PTC reduction effectively converts the 12% bracket into the 22% bracket because of the premium credit phase-out.

So you’d consider whether converting at 22% effective rate is worth it. Compared to deferring and converting later at 22% bracket (with no ACA but with IRMAA concerns) or at 24% bracket once RMDs start: the current conversion may still be the best deal, just less of a slam dunk than it would be without the ACA wrinkle.

Alternative: convert smaller amounts each year. A $20K conversion instead of $50K keeps ACA MAGI lower (~$85K), preserves PTC subsidy at favorable level, still uses 12% bracket. Multi-year strategy of $20K-$30K annual conversions builds Roth over time at lower combined cost.

Get a year-by-year projection done with your CPA. This is exactly the kind of multi-variable optimization that benefits from a spreadsheet projection over 10+ years. Our retirement transition planning handles this routinely.

Your CPA’s number makes sense. Here’s the math:

Gross self-employment income: $180,000. Net self-employment earnings (after deducting half of self-employment tax): $180,000 – $12,718 (estimated SE tax half-deduction at 15.3% × ~$180K × 92.35%) = $167,282. SEP-IRA contribution: $35,000. AGI (line 11): $167,282 – $35,000 = $132,282.

Wait, that gives AGI of $132K, but your CPA said MAGI is $145K. Where’s the difference?

MAGI for Roth IRA includes add-backs to AGI: – Traditional IRA deduction (if any): you didn’t mention any traditional IRA contribution – Student loan interest deduction (if any): didn’t mention – Foreign earned income exclusion: didn’t mention – Other minor items

If those are all zero, MAGI = AGI = $132K, not $145K. Let me try another interpretation.

Possibility: your CPA calculated MAGI to also add back the SEP-IRA contribution. This would NOT be correct for Roth IRA phase-out purposes — the Roth IRA MAGI definition adds back the TRADITIONAL IRA deduction (not the SEP), the student loan interest deduction (not retirement plan contributions), and foreign exclusions. The SEP-IRA deduction reduces AGI and stays reduced for Roth MAGI purposes.

If your CPA added back the SEP-IRA: that’s incorrect. The Roth IRA MAGI calculation under IRC §408A(c)(3) does not include SEP-IRA contributions as an add-back.

Let me try another interpretation. Maybe your CPA’s number of $145K is for a different MAGI — perhaps NIIT MAGI or another rule. If your gross SE income is $180K and you contributed $35K to SEP, your IRA deduction-equivalent MAGI calculation should give you closer to $132K, not $145K.

The other possibility: your CPA correctly applied the Roth MAGI definition but included items I’m not seeing in your summary. Maybe you have tax-exempt interest of $13K that he’s adding back (but tax-exempt interest is added back for ACA MAGI and IRMAA MAGI, not Roth IRA MAGI). Maybe a foreign earned income exclusion. Maybe a traditional IRA deduction that you didn’t mention.

Ask your CPA to walk through the calculation:

‘Can you show me how you got from gross SE income of $180K to MAGI of $145K for Roth IRA purposes? What add-backs did you include?’

If his answer reveals an error (like adding back the SEP-IRA contribution), you might be eligible for a higher Roth contribution than he estimated. If you’re using MFJ filing and his $145K MAGI is correct, you’re below the $236K MFJ phase-out start — full $7,000 Roth contribution allowed. If you’re single, the phase-out is $150K-$165K, and $145K is below start — full contribution. So in either case you should be able to make full Roth contribution.

If MAGI were actually $132K (per my calculation), even more clear you’re below thresholds.

So regardless of which calculation is correct, you appear to be eligible for full Roth contribution. But you should still get clarity on the MAGI methodology because:

1. It affects future planning if you’re near phase-out thresholds.

2. It tells you whether your CPA is applying Roth-specific MAGI vs. some other definition.

3. Other MAGI-based limits (NIIT, ACA, traditional IRA) may yield different numbers, and if he’s confused on Roth, he may be confused elsewhere.

On the SEP-IRA strategically: for self-employed taxpayers, the SEP-IRA contribution limit is 25% of net SE earnings (after the SE tax half-deduction), up to $69K for 2025 (subject to 2026 update). At $167K net SE earnings, 25% = $41,820. You contributed $35K. You could contribute up to about $42K. The additional contribution would lower your AGI further, lower your MAGI, and provide additional tax-deferred retirement savings. Worth considering for the current tax year if cash allows.

Alternative for self-employed: solo 401(k). Allows employee + employer contributions, with employee portion of $23,500 (2025) plus 25% employer portion of net SE earnings. Total contribution can exceed SEP-IRA at the same income level. For high-income self-employed, a cash balance plan layered on top can push retirement contributions into the $100K-$300K range annually.

One more thought: if you’re a high-income freelancer planning long-term, consider whether you want all retirement savings going to tax-deferred (SEP, solo 401k, traditional IRA) or some portion to Roth (post-tax). Tax-deferred is better if your future bracket will be lower than current; Roth is better if future bracket will be higher. For a $180K freelancer in their 30s-40s expecting income growth and possibly state tax changes, a mix is often best. Roth IRA contributions of $7K/year (if under MAGI threshold) combined with traditional SEP contributions of $35K-$42K balances tax diversification.

Real concern. ACA PTC is sensitive to MAGI spikes, and a $400K long-term capital gain will dramatically increase your MAGI for that year. Let’s analyze.

Your current situation (assumed): some income from various sources putting your household at, say, $50K MAGI for ACA. PTC eligibility at maybe 250% of FPL for household of 2 (FPL ~$21K, so 250% = $52.5K). PTC contribution cap might be around 4-5% of $50K = $2K-$2.5K of premium, with the rest covered by PTC. Annual PTC value: probably $8K-$15K depending on actual premium.

Add $400K of long-term capital gain in one year: AGI jumps by $400K. ACA MAGI now $50K + $400K = $450K. For a household of 2 with $450K MAGI, you’re at about 2,140% of FPL — far above any normal subsidy range.

Under the IRA extension (active through 2025): there’s no hard income cliff. The applicable percentage caps at 8.5% of household income. PTC = unsubsidized premium minus 8.5% of $450K = unsubsidized premium minus $38,250. If your unsubsidized premium is, say, $20K/year, then PTC = $20K – $38K = $0 (you owe more than the premium under the formula, so PTC is zero). You’d pay the full unsubsidized premium.

After 2025 (if Congress doesn’t extend the IRA expansion): the original ACA cliff at 400% FPL reappears. Above 400% FPL, no PTC available at all.

Bottom line for 2026 (if IRA expansion expires): selling the Bitcoin in 2026 with $400K of gain pushes you above 400% FPL, eliminating PTC entirely. You’d pay the full unsubsidized premium for the year. Estimated cost: $15K-$25K of premium increase, depending on your area’s premium levels.

For 2026 (if IRA expansion extends): PTC is reduced based on 8.5% formula, probably reducing PTC by $10K-$15K vs. baseline year.

Either way, the year of sale costs you a significant chunk of ACA subsidy. Plus the federal capital gains tax at 20% top rate (if MAGI above ~$553K MFJ threshold; otherwise 15%) plus 3.8% NIIT on the gain. Plus state and city tax.

Let’s run the numbers on a $400K gain:

Federal capital gain at 20%: $80,000 Federal NIIT at 3.8%: $15,200 (since your MAGI is above $250K MFJ NIIT threshold) NYS at 8.82% top rate on gain: $35,280 NYC at 3.876%: $15,504

Total federal + state tax on the sale: $145,984. About 36.5% effective tax.

Plus ACA PTC reduction: $10K-$25K depending on policy environment.

Total effective cost on a $400K gain: maybe 38-42% of gain.

Planning to optimize:

1. Split the sale across multiple years. Sell $100K worth in each of 4 years instead of $400K all at once. Each year’s $100K gain still increases MAGI but stays in a lower bracket and may keep you below the PTC cliff in years that hit the cliff. Federal capital gain rate stays at 15% on most of the gain (the 20% bracket triggers above ~$553K MFJ MAGI). State/city tax brackets are also more favorable at lower income levels. PTC reduction is smaller in each year. Total tax paid over 4 years is likely $20K-$30K less than the single-year sale.

2. Time the sale to align with Medicare eligibility. If you’re approaching 65, hold until on Medicare (no more ACA constraint). IRMAA still applies based on lookback, but the structure is different.

3. Time the sale to align with a low-income year. If you’re approaching retirement and current earnings will drop sharply next year (e.g., retiring December 2026), waiting until 2027 means selling in a year with low base income — the gain pushes you up but from a lower starting point.

4. Loss harvesting offset. If you have other investments with losses (other crypto positions, stock losses, etc.), realize losses in the same year as the Bitcoin gain. Losses offset gains dollar-for-dollar. A $50K net loss harvested reduces the Bitcoin gain to $350K, saving roughly $15K-$20K of combined tax.

5. Charitable giving with appreciated crypto. Donate Bitcoin (or other appreciated assets) directly to a Donor Advised Fund or charity. You get a deduction for fair market value (subject to 30% AGI limit for appreciated property to public charity), avoid the gain entirely, and the charity receives the FMV. For a $400K gain, donating $50K-$100K of the holding shifts that amount entirely out of gain calculation. Charitable bunching can multiply this effect.

6. Section 1031 not available for crypto. Pre-2018 case law sometimes suggested 1031 for crypto, but TCJA explicitly limited 1031 to real property. Crypto-to-crypto exchanges are taxable.

7. Qualified Opportunity Zone investment. If you invest the gain in a QOZ fund within 180 days of sale, you can defer the gain until 2027 (or 7 years from investment, whichever is earlier) and potentially exclude 10-15% of the gain depending on holding period. Caveat: the QOZ deferred gain becomes payable in 2027 unless Congress extends. Watch the rules. This could be useful for timing-shifting the gain to a later year.

For your specific scenario, breakeven analysis:

– Sell now (assumed 2026): full ACA hit + 20% federal rate + 3.8% NIIT + state tax = ~38-42% effective rate. ACA cost: $15K-$25K. Total: $165K-$185K of tax+ACA on $400K gain.

– Spread over 4 years: lower federal rate (mostly 15%, not 20%), smaller annual ACA hits, total tax+ACA maybe $130K-$155K.

– Difference: maybe $35K-$50K saved by spreading.

My recommendation: don’t sell the whole position in one year unless cash flow demands it. Split into 3-4 tranches across years. Consider harvest-offset with any loss positions. Donate a portion to charity if you have charitable intent. Get a CPA’s projection of after-tax proceeds under each scenario before deciding.