Start here
Has newer research replaced the 4% rule with one higher withdrawal rate?
No single updated rate applies to every retirement. Bengen’s later historical work, Morningstar’s forward-looking simulations and flexible spending systems answer different questions. Their figures depend on investment mix, horizon, inflation adjustments, fees, taxes and the definition of success. A larger initial percentage may require accepting smaller later payments or different risks. Canadian readers need those assumptions before using any published number to estimate a portfolio target or retirement spending allowance.
What you’ll learn
- Compare the spending promise and research method before comparing percentages.
- A better average historical outcome need not improve the worst historical case.
- Published withdrawal results require separate scrutiny of costs, taxes and the stability of later payments.
There is no single updated number
A newer withdrawal figure is meaningful only with the question it answers. A historical test can search for an amount that survived the worst observed starting date. A forward-looking model can search for an amount that succeeded in a specified proportion of simulated futures. A flexible-spending method can allow higher initial withdrawals because later payments may fall.
Those are different contracts with uncertainty. Replacing 4% with the largest published percentage combines their apparent benefits while discarding their conditions. The better comparison asks what changed in the portfolio, evidence, horizon, spending promise and definition of success.
This article distinguishes primary research available on October 28, 2026. It does not describe a hypothetical year-end 2026 study or select a rate for an individual household.
What Bengen's later historical work changed
Bengen's September 18, 2025 report compared fixed allocations of 55% stocks, 40% bonds and 5% cash with 65%, 20% and 15%. Across his historical retirement cases, the worst-case initial rates were approximately 4.68% and 4.65%, respectively—both commonly rounded to 4.7%. A larger stock allocation improved many cases without raising that worst-case result. Bengen's September 2025 research report
His accompanying scenario register identifies these as 30-year, inflation-adjusted, tax-advantaged scenarios with no required legacy. They belong to his US historical research framework, not a Canadian-dollar simulation. Its tax-advantaged account assumption does not make a Canadian household’s eventual withdrawals tax-free. Bengen's scenario definitions
This distinction is easy to miss: better average outcomes and a higher minimum sustainable outcome are different findings. An approach can improve most historical retirements while doing slightly worse in the hardest one.
What Morningstar's figure for new retirees means
Morningstar's December 3, 2025 report estimated a 3.9% starting withdrawal for its US-dollar, forward-looking model: constant inflation-adjusted withdrawals, 30 years, and 90% simulated success. The highest base-case rate occurred with 30%–50% equities, the remainder bonds and cash. Its 1,000-trial model excluded taxes and fees; success required funding scheduled withdrawals with assets remaining at the horizon. It was not a Canadian pension-inclusive result. Morningstar: The State of Retirement Income, 2025
The report's publication year matters. This is the 2025 research commonly discussed for people starting retirement in 2026, not a new 2026 historical guarantee. Its estimated returns and inflation differ from simply replaying past market sequences. A difference from Bengen's result is therefore not, by itself, evidence that one calculation contains an arithmetic mistake.
Expanded evidence still has implementation limits
In a March 29, 2026 direct interview, Bengen discussed expanding his historical portfolio beyond the original stock and bond categories. The discussion connects that broader set of assets with the later 4.7% result. Crucially, he states that his research assumes essentially zero investment costs. These remain historical US results, not demonstrated Canadian after-tax spending capacity. Bengen’s 2026 direct interview
An index return and a household’s realized return are different quantities. Fund expenses, advice fees, trading costs and taxes can affect what remains available. A headline is not an allowance to ignore those deductions, and adding asset categories does not guarantee that their historical relationships will repeat.
Spending design matters as well. Holding dollar payments constant transfers inflation risk to the household’s purchasing power. Maintaining purchasing power puts more pressure on the portfolio. A claim that a method permits more spending needs a time dimension: more in the first year, more in real terms over a lifetime, or more only when investment outcomes cooperate? Those claims are not interchangeable.
Fictional CAD example: identical capital, different questions
Assume an $850,000 portfolio and calculate only the initial gross withdrawal. No tax or investment fee is deducted in this first table, and no investment performance is simulated. The percentages are comparison inputs, not conclusions that any of them is suitable.
| Illustrative starting input | Arithmetic | First withdrawal |
|---|---|---|
| 3.9% | $850,000 × 0.039 | $33,150 |
| 4.0% | $850,000 × 0.040 | $34,000 |
| 4.7% | $850,000 × 0.047 | $39,950 |
The difference between 3.9% and 4.7% is $6,800 annually. Relative to $33,150, that is about 20.5% more initial money. The arithmetic is certain; the ability to sustain the larger withdrawal is not established by the table.
If all withdrawals were subject to an invented 20% effective tax, spendable amounts would be $26,520, $27,200 and $31,960. Those tax assumptions are deliberately simplified. The example shows why a before-tax research number and an after-tax household budget cannot be compared directly.
Reverse the calculation without confusing it with proof
Suppose the desired gross portfolio withdrawal is $40,000. Dividing by a chosen percentage gives a screening target: $40,000 ÷ 0.039 is approximately $1,025,641, while $40,000 ÷ 0.047 is approximately $851,064.
The roughly $174,577 difference can strongly influence a retirement date. Yet nothing about division validates the rate. Using a more generous input has simply relaxed the savings requirement mathematically. It has not supplied evidence about future returns, Canadian taxes, the household's horizon or its willingness to reduce spending.
This is why a target calculation needs a separate assumptions page. Otherwise a change in research interpretation can look like an immediate increase in financial capacity, even though no money, pension entitlement or spending flexibility has changed.
A changed-assumption test: the same dollars buy less
Consider a fixed nominal annual withdrawal amount of $40,000. Assume 2.5% annual inflation and evaluate its purchasing power 20 years after the initial date. That $40,000 then buys approximately $24,411 in initial-year purchasing power: $40,000 ÷ 1.025 raised to 20.
Maintaining the original purchasing power would instead require approximately $65,545 at that date. Both schedules can begin at $40,000, but they are funding substantially different later lifestyles.
A rate comparison that does not identify which schedule it uses is incomplete. The same problem arises when a flexible policy allows cuts that a fixed-real policy forbids. A higher starting number may be compensation for accepting a less stable spending path.
Read updates as changes in evidence and assumptions
A useful reading sheet records publication date, historical or simulated data, currency, investment mix, horizon, withdrawal timing, inflation convention, costs, taxes and what counts as failure. Mark a missing item as unknown rather than filling it with a favourable assumption.
For a learning exercise, compare two fictional claims: one survives every observed 30-year historical period; another funds 90% of 30-year simulations. Neither sentence establishes which is safer for the next Canadian household. The answer requires the data, return assumptions and spending policy behind each result.
The practical achievement is understanding what a revised figure purchases and what risk remains. A withdrawal update can improve a model without creating a universal permission to spend more.
Evidence behind the explanation
Sources & checks
Linked sources were consulted on October 28, 2026. The examples are original educational calculations. Source consultation and arithmetic checks do not constitute independent professional review.
- Research Report 1: Optimizing Fixed Asset Allocation During Retirement (September 18, 2025)William P. Bengen · Accessed 2026-10-28
Compared allocations, historical worst-case 4.68% and 4.65%, and distinction between average improvement and worst-case result.
- Charts & Tools for You: scenario definitionsWilliam P. Bengen · Accessed 2026-10-28
Scenarios 2.7 and 2.13: 30 years, COLA withdrawals, tax-advantaged accounts and zero legacy target.
- The State of Retirement Income: 2025 (December 3, 2025)Morningstar · Accessed 2026-10-28
3.9% base case, 30-year horizon, 90% simulated success, allocation, costs/taxes excluded, forward-looking assumptions.
- You Can Spend More in Retirement with Bill Bengen: Bogleheads on Investing Episode 92 (March 29, 2026)John C. Bogle Center for Financial Literacy; direct author interview · Accessed 2026-10-28
Direct author discussion of expanded historical asset classes and the explicit assumption of essentially zero investment costs.
When this article needs another review
- Bengen publishes a new result with complete scenario definitions.
- Morningstar releases a new annual retirement-income study.
- Clarification of fee or data-tail assumptions changes interpretation.
Check the linked official guidance before applying a rule that may have changed.
About this educational article
Published by North of Numbers and created with AI assistance using the sources listed above. This article has not been reviewed by a qualified financial professional. Sources, assumptions and hypothetical examples are shown so readers can examine the reasoning.
This article explains general concepts; it does not assess individual suitability or provide personal financial, investment, tax, insurance, mortgage or legal advice.