Building a Paycheck for Life: Retirement Income Done Right

Trading a paycheck for a portfolio is one of the hardest transitions in personal finance. For decades the money arrived on a schedule and the only question was what to do with it. In retirement you decide when it arrives, where it comes from, and how much — and each choice carries a tax consequence the paycheck never did.

Most attention goes to the first question: how much can I take? Morningstar's most recent modeling puts the safe starting rate near 3.9% for a rigid, inflation-adjusted paycheck across thirty years, at a 90% probability of success — simulation output resting on forward-looking assumptions, not an observed result, and a figure that has moved between 3.3% and 4.0% across recent editions. Retirees accepting substantially variable spending can start above 5%.

Where the money comes from, and in what order, gets far less attention. That is where planning earns its keep.

The default order, and why it is only a starting point

The conventional rule is familiar: spend taxable accounts first, tax-deferred next, Roth last. The logic is sound — a taxable account leaks return to tax every year, while sheltered accounts compound without that drag.

But the rule optimizes for account type while ignoring tax bracket, and that omission is expensive. Followed strictly, it produces long stretches of artificially low taxable income in early retirement, then a wall of required distributions later — income you did not choose, arriving in brackets you did not plan for. A 2015 study in the Financial Analysts Journal concluded the conventional wisdom is wrong: strategies blending withdrawals across account types, including partial Roth conversions, added more than three years of modeled portfolio longevity against the conventional order.

One qualifier is usually left out. That study ran under pre-2018 tax law, and much of the research quantifying these gains assumed the 2017 brackets would expire after 2025 — an assumption Congress has since cancelled. The direction of the finding holds; its size has not been re-established under current law.

The window

Between the year earned income stops and the year required distributions begin, most retirees control their taxable income almost completely. That is unusual, and it expires on schedule.

Under SECURE 2.0, required distributions begin at 73 for those born from 1951 through 1958, and at 75 for those born in 1960 or later. Anyone born in 1959 sits in a genuine drafting gap — the final regulations expressly reserve that year, and proposed guidance would set it at 73, but it is not settled. For someone who stops working in their early sixties, the window now runs eight to thirteen years. Before the SECURE Acts moved the age, it was five to eight.

For 2026, a married couple filing jointly has a $32,200 standard deduction, plus $1,650 each once both are 65. Through 2028 there is also a temporary senior deduction of $6,000 per qualifying individual, reduced by 6% of modified adjusted gross income above $150,000 on a joint return — so a couple both 65 or older loses it entirely by $250,000, and faces a hidden additional 12% marginal rate inside that band. Above the deductions, the 12% bracket runs to $100,800 of taxable income and the 0% rate on long-term capital gains to $98,900.

That room has three uses: converting portions of a traditional IRA to a Roth, realizing long-term gains at 0%, or taking IRA distributions to fill low brackets rather than deferring everything into a larger future problem.

The 0% rate invites one caution. That $98,900 is a taxable income ceiling the gains themselves count toward, stacking on top of ordinary income — so usable room is the ceiling minus everything else. And realized gains raise adjusted gross income, which drives both thresholds described below. A gain harvested at 0% federal can still be paid for twice.

What changed in 2025 — and what did not

Much of the Roth conversion advice written between 2023 and 2025 rested on a deadline: convert before the brackets revert at the end of 2025. That deadline no longer exists. Legislation enacted in July 2025 removed the sunset and made the current rate structure permanent. Conversions did not stop working — but the case for one now has to be made on your own bracket arithmetic rather than on a calendar.

Conversions do not always pay, and since 2017 they cannot be undone. They are close to a wash when today's rate matches the rate you would eventually pay and the tax comes out of the converted funds. They tend to lose when executed in the top brackets by someone whose income is about to fall, when the money was destined for charity and would have escaped tax entirely, or when today's 22% is traded for a future 12%.

The near-universal advice to pay the conversion tax from outside funds carries a condition almost always dropped: it assumes those funds are cash or carry full basis. If appreciated securities must be sold to raise the tax, the resulting gains reduce the conversion's advantage and should be modeled explicitly.

Two thresholds that do quiet damage

Medicare surcharges. The standard Part B premium for 2026 is $202.90 a month. Above $218,000 of modified adjusted gross income on a joint return — a figure that includes tax-exempt municipal interest — a surcharge applies to both Part B and Part D. The lookback is two years, so 2026 premiums are set by the 2024 return, and income recognized at 63 determines what you pay in your first Medicare year. These tiers are cliffs, not phase-ins: one dollar over the first threshold costs roughly $1,150 a year per enrolled person, about $2,300 for a couple both on Medicare. A conversion sized without checking that line can cost more in premiums than it saves in tax.

When income falls because of retirement itself, Form SSA-44 allows a request for a new determination based on current income; work stoppage is a listed life-changing event, and this is a new decision rather than an appeal. It does not reach a surcharge triggered by a Roth conversion — voluntary income recognition is not a qualifying event.

Social Security taxation. Up to 85% of benefits become taxable once provisional income exceeds $44,000 on a joint return; the first tier begins at $32,000. Those thresholds carry no inflation indexing and have not moved since they were written. Within a certain range, each additional dollar of income pulls benefit dollars into taxable income alongside it, producing effective marginal rates well above the stated bracket.

Worth saying plainly, because it has been widely misreported: the 2025 senior deduction did not make Social Security benefits tax-free. It is a separate deduction, it phases out, and it expires after 2028. The rules governing how benefits are taxed were not changed.

Claiming, and the survivor

Claiming Social Security is the largest irreversible decision in most retirement plans. For anyone born in 1960 or later, full retirement age is 67. Claiming at 62 permanently reduces the benefit to 70% of the full amount; waiting until 70 raises it to 124%. That increase is adjusted for inflation and continues for life.

For married couples the higher earner's decision does double duty — though the arithmetic is more forgiving than usually described. A surviving spouse who has reached their own full retirement age generally receives 100% of the deceased's benefit, including delayed credits earned. If the higher earner claimed early, the survivor's amount is limited — but not to the reduced figure. The floor is 82.5% of the deceased's full benefit. Delay still meaningfully raises the survivor's income; it simply does not carry the cliff often described.

Delay also makes the tax window work. Years without benefit income are years with lower provisional income, which is exactly when conversions and gain harvesting cost the least.

De-risking without giving up growth

The specific danger in early retirement is not a bad market. It is a bad market while you are selling, which locks in losses a still-accumulating investor would simply wait out.

One widely cited analysis of US historical data found that inflation-adjusted returns over the first nine or ten years of retirement correlate about 0.81 with the withdrawal rate a portfolio ultimately sustains — against about 0.21 for the first year alone. It takes a bad decade, not a bad year. That argues for holding enough short-duration, high-quality fixed income to fund several years of withdrawals, so spending never forces an equity sale at the wrong moment. That is the job the bond allocation is doing; yield is secondary.

The opposite error deserves equal weight. A thirty-year retirement is an inflation problem as much as a market problem, and a portfolio de-risked until it cannot outpace inflation has exchanged a visible risk for a slower and more certain one.

The work is annual

Every December poses the same question in a new form: how much income should be recognized this year, and from where? The answer depends on this year's brackets and thresholds, what the portfolio did, when benefits begin, and how many years remain in the window. The value is in deciding deliberately rather than by default.