Safe Withdrawal Strategies

Saving is the easy half. The hard half is turning a pile of money into a paycheck that lasts as long as you do, without either running out or dying with a fortune you never spent. This guide covers the 4% rule, what it actually claimed, why sequence of returns is the real enemy, and the spending rules that adapt.

By Karanvir Singh, founder of Quantix Invest

The decumulation problem

While you are working, the math of investing is forgiving. You add money every month, you never sell, and a bad decade is an opportunity rather than a wound. Once you retire, every one of those cushions disappears. You sell instead of buy, the contributions stop, and a bad decade at the wrong moment can permanently shrink what you have to live on.

The question is deceptively simple: given a portfolio of some size, how much can you take out each year with reasonable confidence you will not run out before you run out of years? Three unknowns collide.

• How long you will live. A 65-year-old US couple today has a meaningful chance that at least one of them reaches 95. Planning for the average lifespan means roughly half of people outlive the plan.
• What markets will do. Not just the average return, but the order in which the returns arrive, which matters enormously once you are selling.
• What you will actually spend. Real retirement spending is lumpy, not a smooth inflation-adjusted line: higher early, lower in the middle, sometimes sharply higher at the end for health and care costs.

Every withdrawal strategy is an attempt to manage those three unknowns. None of them removes uncertainty. The good ones simply make the failure modes visible and recoverable.

Where the 4% rule came from

In 1994, a financial planner named William Bengen published a study asking a specific historical question: for every 30-year retirement window in US market history, what is the highest starting withdrawal rate that would have survived the worst of them, if you raised the dollar amount each year for inflation and held a stock-heavy balanced portfolio?

His answer was roughly 4%. A few years later, the Trinity study (three professors at Trinity University) ran a similar analysis on success rates across stock and bond mixes and reached compatible conclusions, and the phrase "the 4% rule" entered the language.

The mechanics of the rule as tested:

• Year one, withdraw 4% of the starting portfolio value. On $1,000,000 that is $40,000.
• Every year after, increase the dollar amount by inflation, ignoring the portfolio entirely. If inflation runs 3%, year two is $41,200, year three about $42,436, and so on.
• Hold something like 50% to 75% stocks with the rest in bonds, rebalanced.
• Success means the money lasted 30 years.

The result is a useful anchor because it was calibrated against genuine disasters. The worst historical starting points, retiring into the 1929 crash, into the 1937 relapse, and above all into the mid-to-late 1960s inflation era, are what set the number. In the median historical case, a 4% start left the retiree with more money after 30 years than they began with, often several times more.

What the 4% rule does not say

Most arguments about the rule are arguments with things it never claimed.

• It is not "4% of the current balance." Taking 4% of whatever the portfolio is worth each year is a completely different rule. That version can never run out (4% of a shrinking number is always positive) but the income swings wildly with the market. The Bengen rule fixes spending in real terms and lets the portfolio absorb the volatility.
• It is not a guarantee. It is a historical worst case for one country, over overlapping windows drawn from a single long sample, in an era of unusually strong US returns.
• It is not a 60-year rule. The test horizon was 30 years. Early retirees planning for 45 or 50 years need a lower starting rate, in the neighborhood of 3% to 3.5% by most analyses, because a longer horizon has more chances to encounter a ruinous sequence.
• It says nothing about taxes or fees. A 4% withdrawal from a traditional IRA is not 4% of spendable income; income tax comes out of it. And a 1% advisory fee plus 0.6% fund costs is, in effect, a permanent extra 1.6% withdrawal that never adjusts downward in a bad year.
• It assumes you behave. The historical success rates assume you rebalanced into stocks during the crash instead of selling out.

Bengen himself later revised upward, arguing that with broader diversification (adding small-cap and other asset classes) the historically safe rate was closer to 4.5%, and in more recent work he has floated higher figures still for the median case. Other researchers went the opposite direction. That disagreement is the honest state of the evidence.

Sequence-of-returns risk: a worked example

Here is the mechanism that makes retirement different from saving. Two retirees, identical portfolios, identical average returns, wildly different outcomes, because of the order the returns arrive in.

Both start with $1,000,000 and withdraw $40,000 at the end of each year, ignoring inflation for simplicity. Both experience the same three returns over three years: minus 20%, plus 10%, plus 30%. Retiree A gets them in that order. Retiree B gets the reverse: plus 30%, plus 10%, minus 20%.

Retiree A (bad years first):

• Year 1: $1,000,000 falls 20% to $800,000, withdraw $40,000, ending balance $760,000.
• Year 2: $760,000 grows 10% to $836,000, withdraw $40,000, ending balance $796,000.
• Year 3: $796,000 grows 30% to $1,034,800, withdraw $40,000, ending balance $994,800.

Retiree B (good years first):

• Year 1: $1,000,000 grows 30% to $1,300,000, withdraw $40,000, ending balance $1,260,000.
• Year 2: $1,260,000 grows 10% to $1,386,000, withdraw $40,000, ending balance $1,346,000.
• Year 3: $1,346,000 falls 20% to $1,076,800, withdraw $40,000, ending balance $1,036,800.

Same three returns, same withdrawals, and B ends with $42,000 more after only three years. Stretch the same effect over a 30-year retirement with inflation-adjusted withdrawals and the gap becomes the difference between dying with millions and running out at 84.

Why does this happen? Because selling during a decline converts a paper loss into a permanent one. Retiree A's year-one $40,000 came out of a portfolio that had already fallen. Those shares are gone and cannot participate in the recovery. A saver in the accumulation phase experiences the mirror image: a crash early in your career is a gift, because every contribution buys cheap shares.

The critics, and what they get right

Three serious lines of criticism, in rough order of importance.

1. Valuations and yields matter. The historical record includes many retirements that began with cheap stocks and high bond yields. Research by Wade Pfau, Michael Kitces and others shows a clear relationship: retirements beginning at high equity valuations (a high cyclically adjusted P/E) had materially lower safe withdrawal rates than those beginning at low valuations. Starting from expensive markets, a more conservative initial rate is the prudent reading.

2. US survivorship bias. The 4% figure rests on the twentieth-century returns of the most successful market in the world. Studies extending the analysis to a broad panel of developed countries (work by Anarkulova, Cederburg and O'Doherty among others) find substantially lower safe rates once you admit that an investor could have been Japanese, German, or Italian rather than American. A globally diversified stock allocation partly answers this critique, but it does not make it disappear.

3. Nobody actually spends that way. The rule assumes a robotic inflation-adjusted withdrawal regardless of what markets do. Real retirees cut back when their balance falls, which is precisely why static-rule failure rates overstate real-world failure. Spending studies also consistently find real spending declines with age through the sixties, seventies and early eighties, before rising again for health and long-term care. The "smile" shape means a flat inflation-adjusted plan is conservative in the middle and possibly thin at the end.

What the critics do not establish is that the rule is useless. As a planning anchor, "multiply your desired annual spending by 25 and that is roughly your target" remains the fastest useful sanity check in personal finance. Treat 4% as the starting hypothesis, then adjust for horizon, valuations, fees, and flexibility.

Guardrails and dynamic spending

If rigid spending is the problem, rules that flex are the answer. The best known is the guardrails approach developed by Jonathan Guyton and William Klinger. The idea: set an initial rate, then define upper and lower boundaries on your current withdrawal rate that trigger a spending change.

A common implementation:

• Start at, say, 5% of the portfolio, higher than the static rule because you have agreed to adjust.
• Each year, adjust the dollar amount for inflation, then compute the withdrawal as a percentage of the current balance.
• If that percentage rises above the upper guardrail (often 20% above the starting rate, so 6.0% if you started at 5%), cut spending by 10%.
• If it falls below the lower guardrail (often 20% below, so 4.0%), raise spending by 10%.
• Skip the inflation raise in any year following a negative portfolio return.

Worked example. Start with $1,000,000 and a 5% rate: $50,000 in year one, leaving $950,000. Suppose the market falls hard and the portfolio ends year two at $720,000 while inflation of 3% would push spending to $51,500. That is 51,500 / 720,000 = 7.15%, well above the 6.0% upper guardrail. The rule triggers: cut spending 10%, to $46,350, and skip the inflation raise. The new rate is 46,350 / 720,000 = 6.4%. Still high, but a $5,150 reduction in one year has meaningfully extended how long the portfolio can carry you, and the cut is a fraction of what a panicked mid-crash reaction usually looks like.

The tradeoff is honest and worth stating plainly: guardrails let you start higher, at the price of accepting real spending cuts of 10% or so, occasionally more than once, in bad markets. That is only a good trade if a meaningful share of your budget is genuinely discretionary. If nearly all of your spending is fixed (housing, insurance, food, medicine), you cannot flex, and you should start lower rather than pretend you can cut.

Variable percentage withdrawal

A cleaner cousin of guardrails is the variable percentage withdrawal approach: each year, withdraw a percentage of the current balance where the percentage rises with age, roughly on an amortization schedule over your remaining expected years.

Illustrative percentages for a balanced portfolio look something like 4.0% in the mid-sixties, near 5% at 75, near 7% at 85, and higher still past 90. The reason the percentage climbs is simply that the money has fewer years left to cover.

Properties worth knowing:

• It cannot fail. A percentage of a positive balance is always positive, so you never hit zero.
• The income does swing. A 30% market drop cuts the withdrawal by roughly 30% that year. This is the honest cost of never running out.
• It spends more in good markets. Static rules systematically underspend, since in most historical cases the retiree died with a far larger portfolio than they started with. VPW recycles that surplus into your own lifetime instead.

The practical version pairs VPW with a guaranteed floor from Social Security and possibly an annuity, so that when the variable part shrinks, essentials are still covered.

Bucket strategies and cash reserves

The bucket approach segments the portfolio by when you will spend it:

• Bucket 1 (roughly 1 to 3 years of spending): cash, money market funds, T-bills, short CDs. This is what you actually live on.
• Bucket 2 (roughly years 4 to 10): high-quality intermediate bonds, TIPS, bond funds.
• Bucket 3 (year 10 and beyond): a globally diversified stock portfolio.

You spend from bucket 1, and refill it from buckets 2 and 3, typically during rebalancing and preferentially from whatever has done well. The mechanism is meant to prevent forced stock sales in a downturn: if stocks are down 35%, you have several years of spending sitting in cash and bonds and can simply wait.

Two honest caveats. First, researchers have repeatedly shown that a bucket portfolio is mathematically close to a plain rebalanced allocation with the same overall stock and bond weights. If your buckets add up to 60% stocks, you own a 60/40 portfolio with extra vocabulary. Second, a large permanent cash bucket costs return over decades. What buckets genuinely deliver is behavioral: a retiree who can point at a labeled account holding three years of groceries is far less likely to sell equities at the bottom, and staying invested is worth more than the drag.

Building a floor: Social Security, TIPS ladders, annuities

Withdrawal rates get easier the less of your essential spending they have to cover. The strongest structural move available to most retirees is not a clever spending rule; it is enlarging the guaranteed, inflation-adjusted, lifelong part of their income.

Delaying Social Security. Between the earliest claiming age of 62 and age 70, the monthly benefit grows substantially each year you wait, and the increase is permanent and inflation-adjusted for as long as you live. For a married couple, delaying the higher earner's benefit also raises the survivor benefit, which is exactly the protection the longer-lived spouse needs. Bridging with portfolio withdrawals from 62 to 70 means spending more early so that a guaranteed floor is larger later, which is usually a better trade than the reverse. Individual circumstances (health, work, spousal benefits) genuinely change the answer here, so this is education rather than a recommendation.

TIPS ladders. Buying individual Treasury Inflation-Protected Securities maturing in each of the next 20 or 30 years builds a known, inflation-adjusted stream of principal payments with no market risk and no insurance company involved. The limits: TIPS currently extend to 30 years, so a ladder does not cover longevity beyond its end date, and the annual inflation adjustments create phantom taxable income in taxable accounts, which argues for holding them in an IRA.

Annuities. A single premium immediate annuity converts a lump sum into payments for life. A deferred income annuity purchased at 65 to begin at 80 or 85 is the cheapest form of pure longevity insurance, because you are only paying for the tail. The costs are real: you give up liquidity and any bequest of that money, most annuities are not inflation-adjusted (and the ones that are start much lower), and you take on the insurer's credit risk, mitigated but not eliminated by state guaranty associations. Complex variable and indexed annuities sold with high commissions are a different product entirely and deserve deep skepticism.

Required minimum distributions

Tax-deferred accounts do not stay deferred forever. Required minimum distributions force money out of traditional IRAs, 401(k)s, 403(b)s and similar accounts so the government finally collects its tax.

The key facts as of 2026:

• RMDs begin at age 73 for those who reached 72 after 2022, and move to age 75 for those born in 1960 or later, under the SECURE 2.0 changes.
• The amount is your December 31 prior-year balance divided by a life expectancy factor from the IRS Uniform Lifetime Table. At 75, the factor is 24.6, so the RMD is about 4.07% of the balance. The percentage rises every year.
• Roth IRAs have no RMDs for the original owner. As of 2024, designated Roth accounts inside 401(k) plans also no longer have RMDs during the owner's life.
• Missing an RMD triggers a penalty of 25% of the shortfall, reduced to 10% if corrected promptly. Take them.
• Inherited IRAs generally follow a 10-year rule for most non-spouse beneficiaries, with annual distributions also required in many cases.

Worked example. A 75-year-old with $900,000 in a traditional IRA has an RMD of 900,000 / 24.6 = $36,585. Suppose their spending plan only needed $30,000 from that account. The extra $6,585 must still be distributed and taxed, though nothing stops them from reinvesting it in a taxable brokerage account. If they are charitably inclined and over 70 and a half, a qualified charitable distribution sends up to an annually indexed limit (over $100,000, adjusted for inflation) straight from the IRA to charity, counting toward the RMD and staying out of taxable income entirely, which is usually better than taking the distribution and claiming a deduction.

The larger planning point: RMDs stacked on top of Social Security can push retirees into higher brackets and trigger Medicare IRMAA surcharges. The low-income years between retirement and the RMD start date are the natural window for Roth conversions, deliberately moving money from traditional to Roth and paying tax at today's lower rate to shrink the future forced distributions.

Which account to spend first

The conventional default ordering is taxable first, then tax-deferred, then Roth last, because it leaves the tax-sheltered compounding running longest and preserves the most flexible account for the end.

The more sophisticated answer is that pure ordering is usually beaten by bracket management: each year, fill up the lower tax brackets with traditional-account withdrawals or Roth conversions, and take anything beyond that from taxable or Roth so you do not spill into a higher bracket. Things to watch alongside the brackets:

• Long-term capital gains have a 0% rate band for lower taxable incomes, making early retirement a chance to realize gains tax-free.
• Medicare IRMAA surcharges are cliffs, not ramps. One dollar over a threshold raises premiums for the whole year, based on income from two years earlier.
• The share of Social Security benefits that is taxable rises with other income, creating awkward effective rates in certain ranges.
• Heirs are affected: appreciated taxable assets currently receive a step-up in cost basis at death, while inherited traditional IRAs arrive fully taxable to the beneficiary.

This is education, not individualized advice, and withdrawal sequencing is one of the places where the details of your own tax situation genuinely change the right answer.

Common mistakes

• Treating 4% as a law of nature. It is a historical worst case for a 30-year US retirement with no fees and no taxes. Your horizon, costs and flexibility all move the number.
• Forgetting fees are a withdrawal. A 1% advisor fee on top of a 4% withdrawal is a 5% drain. Fees do not adjust downward when markets fall.
• Ignoring taxes in the plan. $40,000 out of a traditional IRA is not $40,000 of spending money.
• Retiring 100% in stocks. The sequence example shows why the years around your retirement date are the wrong time to be maximally exposed.
• Retiring 100% in cash and bonds. A 30-year retirement faces roughly a doubling of prices at 2.5% inflation. Without growth assets, safety in the first decade becomes poverty in the third.
• Refusing to ever adjust. The single cheapest source of safety is a pre-agreed list of expenses you will trim after a bad year.
• Overspending in the first two years. The "we finally have time" phase is real, and it lands exactly in the window where sequence risk bites hardest.
• Skipping an RMD. An avoidable penalty on money you were going to be taxed on anyway.
• No plan for long-term care. The end of the spending smile is the largest single uncovered risk in most retirement plans.

Bottom line: start from something near 4% for a 30-year horizon and lower for a longer one, subtract your real costs, build the largest guaranteed floor you reasonably can, and write down in advance what you will cut if the first five years go badly. A plan that bends does not break.