Withdrawal sequencing decides how much of a retirement portfolio survives the tax code

A retiree with three buckets of money faces a question that looks administrative and is not. Taxable brokerage, tax-deferred retirement accounts and tax-free Roth accounts each hand the tax authority a different amount at a different time. Deciding which one funds this year's spending is a decision about where a lifetime of tax liability gets recognised, and the ordering compounds for thirty years. The advice most people receive is to spend the taxable account first, then the tax-deferred account, then the Roth. It has the virtue of being memorable. It maximises the time that tax-advantaged money stays sheltered, which sounds obviously correct. It also guarantees that the tax-deferred balance grows untouched until required distributions force it out in large annual slices at the highest marginal rate the household will ever face.

What each dollar actually costs

Start with the mechanics, because the strategy falls out of them. The three account types differ in when tax is paid, on what, and at what rate.

Two rows do most of the work. The taxable account gets a basis step-up at death, so an embedded gain held until then is never taxed to anyone. The traditional account gets the opposite treatment: it carries a deferred ordinary income liability that an heir must pay, usually compressed into ten years and often during the heir's highest earning decade. Any plan that leaves a large traditional balance to a working-age child has chosen the most expensive possible outcome for that money.

The window that matters

Most households have a period between the end of employment income and the start of required distributions and Social Security when taxable income falls dramatically. It might run from age 62 to 73. Eleven years of low income, during which the lower brackets sit almost empty. Spending exclusively from the taxable account during that window wastes every one of them.

The alternative is bracket filling. Each year, draw enough from the tax-deferred account, or convert enough of it to Roth, to reach the top of a chosen bracket and no further. The household pays tax voluntarily at a known rate now in order to avoid paying at an unknown and probably higher rate later, and it shrinks the balance that will eventually be subject to required distributions.

Three things stand out. Filling the 22% bracket produces the lowest lifetime tax and the largest terminal portfolio in this example, and it does so by paying nearly four times as much tax in the first decade as the conventional approach. Filling the 24% bracket pays too much too early and the benefit reverses. And the conventional taxable-first strategy, which pays the least tax during the window, ends up paying the most across the whole life of the plan.

The thresholds that create cliffs

Marginal rates are the easy part. The harder part is a set of thresholds where crossing by one dollar changes the treatment of many dollars. These do not appear in the bracket table and they routinely turn a sensible conversion into an expensive one.

• Medicare income-related premium adjustments apply in tiers based on income from two years earlier, and crossing a tier boundary by a dollar raises annual premiums for both spouses for a full year.
• The long term capital gains 0% band ends at a specific taxable income figure, and a deferred-account withdrawal that pushes past it can subject previously untaxed gains to 15%.
• The 3.8% net investment income surtax applies above a modified adjusted gross income threshold that has never been indexed to inflation.
• Taxation of Social Security benefits phases in over a range where each additional dollar of other income can make up to 85 cents of benefit taxable, producing effective marginal rates well above the stated bracket.

That last one produces what planners call the tax torpedo. A household in the nominal 12% bracket can face an effective marginal rate near 22% across the phase-in range, which is why bracket filling is usually done before Social Security starts rather than alongside it.

Where the standard advice is right

Taxable-first is the correct answer in a narrower set of cases than its popularity suggests, and those cases are real. A household whose tax-deferred balance is small relative to spending will never face large required distributions, so there is nothing to pre-empt. A household expecting to move from a high tax state to a no-tax state should generally wait. A household intending to leave the entire tax-deferred balance to charity should never convert, because the charity pays no tax either way and converting simply donates money to the tax authority first.

Clients hear 'pay tax you did not have to pay this year' and stop listening. The way through it is to show the required distribution at 75 under both plans. Once someone sees a forced seventy thousand dollar withdrawal they do not need, the conversation changes in about a minute.

Margit Sollers, tax director at Thistledown Advisors

How to run it without a model

A household without access to planning software can get most of the benefit with an annual routine. In November, estimate the year's taxable income before any voluntary withdrawal. Find the top of the target bracket. Subtract. Convert or withdraw that amount from the tax-deferred account, stopping short of the nearest Medicare tier boundary. Pay the tax from the taxable account rather than from the conversion, because paying from the conversion shrinks the sheltered balance and defeats part of the purpose. Do it every year for a decade and the required distribution that arrives at 73 is a fraction of what it would otherwise have been. The portfolio is not bigger because the investments performed better. It is bigger because less of it was ever sent away, and because the timing of what was sent away was chosen rather than imposed.