Domicile and the Conversion Year

Federal tax is uniform across the United States; state tax is not. The same Roth conversion costs you 13.3% more in California than in Texas, and 0–10.9% more in New York depending on the bracket. Over a 10-year conversion window on a $3M pre-tax balance, the state-tax delta between a high-tax and no-tax domicile can run into the high six figures.

The California exit imperative Federal tax is a negotiation; California state tax on the wealthy is closer to confiscation, and the Franchise Tax Board (FTB) pursues departing residents with a demonstrated appetite for litigation and no bright-line day test to hide behind. Clear one rumor first: there is no California exit tax. AB 2088 (2020) proposed 0.4% a year on net worth above $30 million with a ten-year tail that would have kept taxing you on a declining share after you left; AB 259 (2023) tried again at 1% above $50 million and 1.5% above $1 billion. Neither got out of committee. What follows you across the state line is ordinary residency and source law — older, narrower, and enforced with more energy than any legend. The 2026 stack as it applies to a retiree with significant income:

Ordinary-income brackets

top at 12.3% on taxable income above roughly $700,000 single or $1.4M MFJ. A Roth conversion is ordinary income in California.

Behavioral Health Services Tax

(formerly the Mental Health Services Tax, Prop 63) adds 1% on every dollar of taxable income above $1M, taking the combined CA top rate to 13.3%.

No preferential treatment for long-term capital gains

. Federal 0%, 15%, and 20% brackets do not exist in California; every realized gain is taxed at ordinary brackets.

Uncapped SDI on earned income

under SB 951. The 2026 rate is 1.3% on every dollar of W-2 wages with no upper bound. This does not apply to a Roth conversion, an RMD, or a realized gain—it is a payroll tax on earned income—but it materially raises the cost of staying in California while drawing a W-2 consulting paycheck.

Size the bill accurately instead of applying the top rate to the whole conversion. California’s brackets climb, and the 13.3% figure applies only to dollars above roughly $1.44M of MFJ taxable income; the 1% BHST kicks in at $1M, so the band between them runs at 12.3%. Run the schedule on a $2,000,000 conversion as a couple’s only income and it blends out to roughly $219,500 of California tax — an effective state rate of about 11%, not 13.3%. It is the marginal dollars that cost 13.3%, which is exactly why a multi-year ladder run entirely inside California is worse than the headline suggests: each successive year refills the same climbing schedule. The same conversion executed in Nevada, Texas, Florida, or Washington after a clean, properly documented domicile change costs zero. Across a ten-year ladder, this math is the price of a small estate.

The same is true to varying degrees of New York (top rate 10.9% and an aggressive auditor of departing residents) and Hawaii (top rate 11%). Oregon, Minnesota, and New Jersey land below those extremes but well above the no-tax states. If you have already accepted that you will pay federal tax on a large conversion ladder, leaving the state-tax exposure on the table is the financial equivalent of leaving the tip in your seven-figure dinner check.

The choreography of domicile Domicile changes are scrutinized aggressively. The state you are leaving will audit if the numbers are large. Establish domicile in your no-income-tax state before the year of large conversions or realizations. Domicile is a factual determination: voter registration, driver’s license, primary residence, time spent, the location of your physicians and financial advisors, and even your planned burial plot. The two hardest exits are hard in complementary ways: New York runs the most residency audits in the country, while California litigates the closest-connections question hardest and, as the statutory-residency discussion below notes, offers no day-count safe harbor at all.

Time large conversions and realizations into post-move years. A single conversion in the wrong domicile year can pay for the entire cost of the move. Be exceptionally careful with part-year resident filings; a move during a tax year often produces a prorated state liability. A conversion executed before the move date attaches to the prior state. If you plan to retire to a high-tax state for family reasons, front-load conversions in the low-tax origin state in the years immediately before the move.

The snapshot-date problem The strongest argument for moving early is that a legislature can fix the residency date behind you. California’s Proposition 40 on the November 2026 ballot — the “2026 Billionaire Tax Act” — proposes a one-time 5% tax on the net worth of anyone who was a California resident on January 1, 2026, measured as of December 31, 2026, phasing in between $1 billion and $1.1 billion (the rate drops 0.1 point for every $2 million below $1.1 billion), payable in one sum or in five annual installments carrying a 7.5% nondeductible deferral charge, with a ten-year window for the FTB to assess. The threshold is irrelevant to most readers; the structure is not. The measure qualified for the ballot in June 2026, five months after the residency date it looks back to, so nobody who read the news and moved could get out from under it, and its drafters said as much. Every resident who files a 2026 return, billionaire or not, must declare that net assets were at or below $1 billion, and the FTB is directed to examine any declaration it has reason to doubt. Whether a retroactive residency snapshot survives a due-process challenge is a question for the courts; whether the technique gets copied at lower thresholds is a question for the next fiscal crisis. The lesson is evergreen: the year you defer a move is a year a snapshot can land on, and a move completed and documented before the snapshot is the only defense that does not depend on litigation.

The federal shield, and its edges The strategy works because of an explicit federal statute, not an unintended loophole. The Pension Source Tax Act of 1996 (P.L. 104–95, codified at 4 U.S.C. §114) bars any state from taxing the retirement income of a person who is neither a resident nor a domiciliary of that state. The protected list is broad: IRC §401(a) qualified trusts, IRC §403(a) and (b) annuities, IRC §457 plans, simplified employee pensions, governmental plans, military retired pay, and individual retirement accounts. New York cannot reach back and tax the distribution from an IRA funded entirely with New York wages and New York deductions once you are a Floridian. There is no clawback of the deduction you took on the way in, and this is the single fact that makes the entire domicile play work.

The shield has edges. Nonqualified deferred compensation is protected only if it is paid in substantially equal installments over your life expectancy or at least ten years — a lump sum or a five-year payout from a deferred-comp plan remains fully taxable by the state where you earned it. Gain on in-state real property, in-state business and partnership income, and in-state wages stay sourced to that state forever. And nothing in the statute helps a conversion executed while you were still a resident. The protection turns entirely on the word “resident” — a word the departing state gets to define.

Statutory residency: the trap a clean domicile change does not close Domicile is one test; most high-tax states run a second, purely mechanical one alongside it. Maintain a permanent place of abode in the state and spend more than 183 days there, and you are a statutory resident taxed on 100% of your income for that year — the conversion included — no matter that your license, your voter registration, your doctors, and your intent are all in Florida. You can win the domicile argument and still lose the year. New York, New Jersey, Connecticut, Massachusetts, and Pennsylvania all run some version of this. In New York the abode must be maintained for more than ten months of the year to count, and Matter of Obus (206 A.D.3d 1511, 3d Dep’t 2022) held that a genuine vacation house the taxpayer barely used was not a permanent place of abode — but that was won on appeal after years of litigation, which is not a plan. Any part of a day physically present counts as a full day.

The prescription is unsentimental: in any year you convert, sell the old-state house or stay under the day count with real margin — target 120 days instead of 179, because audits are fought one disputed day at a time. California is not friendlier here, only vaguer: it has no bright-line statutory-resident rule, so no day count makes you safe, and R&TC §17016 presumes residency for anyone present more than nine months. The FTB decides by closest connections — the factors catalogued in Appeal of Bragg (2003-SBE-002): where your home, family, license, vote, doctors, advisors, and bank accounts sit, weighed together with no single one controlling — which is exactly the fight you do not want during a seven-figure conversion ladder. The one statutory safe harbor runs the other way and was built for expatriate employees, not retirees: R&TC §17014(d) treats a domiciliary as a nonresident if absent at least 546 consecutive days under an employment-related contract, with no more than 45 days back in the state per year — and it evaporates if your income from stocks, bonds, and other intangibles exceeds $200,000 in the year, or if the move was for tax avoidance. A taxable portfolio large enough to make the ladder worth building trips the $200,000 test on its own.

Day-count evidence A residency audit lives or dies on whether you can show, day by day, where you slept. Both the CA FTB and New York run on the statutory-day test. Reconstructing that calendar two years after the fact from credit-card swipes and calendar entries is a losing exercise. The burden of proof rests entirely on you: the auditor asserts, and you disprove. The auditor also has subpoena power over your phone logs, E-ZPass accounts, and bank statements, and they will use it. Mobile residency trackers—such as Monaeo, TaxBird, or TaxDay—run silently in the background, count days against each jurisdiction’s threshold, alert you before you trip one, and export an audit log your CPA can hand over on demand. The annual subscription is trivial compared to the tax exposure they document.

Price the move net — and check whether you need it Run the conversion schedule through your actual state’s rules before you price moving vans, because the headline top rate is frequently not what you would pay. Pennsylvania’s 3.07% flat tax does not touch a conversion at all; distributions after 59½ are exempt, and the conversion never appears on the PA-40. Illinois levies 4.95% on wages and 0% on IRA distributions, so a conversion there is already free. New York excludes $20,000 per person per year of pension and annuity income once you are 59½, so a couple converting $350,000 is taxed by New York on $310,000. The states where the arithmetic genuinely demands a move are California, New York at the top brackets, New Jersey, Oregon, Minnesota, and Hawaii — not every state with an income tax.

Ignore the retiree-magazine version of this question, which fixates on how each state treats Social Security. As of 2026, forty-two states and the District of Columbia do not tax benefits at all, and the eight that do—Colorado, Connecticut, Minnesota, Montana, New Mexico, Rhode Island, Utah, and Vermont—exempt them under income thresholds or age-based deductions that shelter most retirees anyway. The benefit check is the smallest number on your return and the one you have the least control over. Choose the state on how it taxes conversions, realized gains, and your estate; those are the numbers with commas in them.

Then net the move against what the destination costs. No-income-tax states recover the revenue somewhere: coastal homeowners and flood insurance that can run several times what you paid inland, HOA dues and special assessments on the condo, and a property-tax bill reassessed to your purchase price, not the long-held assessment you enjoyed at home. Add two rounds of transaction costs if the plan is to relocate again later — and many do, drifting partway back toward grandchildren once the conversion ladder is finished. A move you were going to make anyway carries the conversion savings as pure upside. A decade of residency purchased solely to buy the arbitrage nets out thinner than the spreadsheet promised, and it is the only version of this plan that can lose.

The move itself is rarely the goal. But once a move is on the table, the conversion calendar should be built around it. Treat the year of domicile change as the most valuable single-year arbitrage available in retirement planning, and do not waste it.