For years, the 4% rule was the FIRE community's holy grail. Simple, reassuring, precise: withdraw 4% of your portfolio every year, adjusted for inflation, and never run out of money. The Trinity Study had proven it, hundreds of blogs had amplified it, and thousands of aspiring early retirees had built their entire life plans around it.
I believed in it for a while, too. But after 25 years of investing and several years of actual retirement, I can no longer defend this approach in good conscience. Not because it is wrong in absolute terms — but because it rests on foundations that do not survive critical scrutiny.
The core problem: SWR needs data it cannot have
The Safe Withdrawal Rate (SWR) works through historical simulation. You test every possible retirement window in history, identify the worst-case scenario, and set the withdrawal rate to survive even that outcome. It is rigorous. It is also where the problem begins.
To cover a 30-year retirement with statistically meaningful simulation windows, you need data spanning many decades. The original Trinity Study uses 70 years. Extend the retirement duration — as you must for early FIRE — and you must widen the dataset further. This is what serious research like ERN's does, going back to 1871 and concluding, incidentally, that a fixed 4% withdrawal rate is generally too high for retirements beyond 30 years.
What is striking is that the father of the 4% rule, Bill Bengen, moves in the opposite direction. In his 2025 book A Richer Retirement: Supercharging the 4% Rule to Spend More and Enjoy More, he raises the safe withdrawal rate to 4.7% (for a 30-year horizon). Tyler at Portfolio Charts arrives at 4.8%. How do we explain these apparent contradictions?
The answer lies in SWR's own mechanics. Tyler goes back only to 1970. Bengen to 1920. ERN to 1871 — which makes sense since he tests retirements up to 60 years long. Different datasets produce different results. Does this mean ERN's longer series are more correct than Bengen's or Tyler's? Not necessarily. Data since 1871 does not form a coherent whole. It spans radically different economic regimes:
- The strict gold standard (pre-1971), where central banks had none of today's monetary flexibility
- Two world wars and their reconstructions
- Financial repression in the 1940s–50s, when bond yields were administratively set
- The hyperinflation of the 1970s
- The post-Bretton Woods era, quantitative easing, the rise of ETFs, and online trading
As Keynes put it: "In the long run, we are all dead." And that is precisely SWR's dilemma — it needs the very long run to function, but that long run is populated by economic realities that bear little resemblance to the world an early retiree will actually inhabit.
Using data from 1871 to navigate a portfolio in 2025 is like using horse-carriage accident statistics to calculate the safety of a Tesla. The data exist, they are measurable — but they describe a different world.
Rigid withdrawals: a false sense of security
The second major flaw of SWR is architectural. Withdrawals are fixed — inflation-adjusted, but fixed. This rigidity has a direct consequence: to guarantee portfolio survival in the worst historical scenario, the withdrawal rate must be extremely conservative.
The result: in the vast majority of historical scenarios, the portfolio does not merely survive — it explodes. Simulations routinely show portfolios ending with 5, 10, even 20 times their initial value after 30 years of withdrawals. This is not financial security; it is involuntary wealth accumulation. SWR does not produce serene retirees — it produces wealthy heirs.
A retiree who impoverishes their own retirement through excessive caution misses the point entirely: enjoying what they built. Bengen himself acknowledged this. In his 2025 book, he proposes an updated portfolio with a 4.7% withdrawal rate — an implicit admission that his original rule was too conservative. I analyze it in full in this article.
Gold: the collateral victim of a methodological flaw
This temporal bias has concrete victims. Gold is the most revealing one.
Several widely cited analyses in the FIRE community advise against gold, drawing on backtests going back to 1870. They show that gold adds little or nothing to the safe withdrawal rate. Conclusion: gold has no place in a retirement portfolio.
But before 1971, gold was not a market asset. It was an administered price: fixed at $35/oz by Roosevelt's Gold Reserve Act in 1934, then codified internationally by the Bretton Woods agreements in 1944. Its "performance" over that period does not reflect market behavior — it reflects a political decision. Including that data in a backtest means contaminating the analysis with observations that do not measure what they appear to measure.
| Period | Gold status | Real CAGR |
|---|---|---|
| 1934–1971 | Administered price ($35/oz fixed) | ~-2.9% |
| 1972–2026 | Free-market asset | +4.98% |
Since 1971, when Nixon ended dollar-gold convertibility and freed gold to market forces, the picture is radically different. Over the period January 1972 – February 2026, gold's nominal CAGR reaches 9.07% in USD, against 10.84% for the US equity market. In real terms (inflation-adjusted), gold delivers 4.98% annualized against 6.68% for equities — consistent with Jeremy Siegel's data for the post-1971 period. The gap exists — equities remain superior over the long run, which makes sense since companies create value — but gold is no longer the sterile, unproductive asset that long-horizon backtests portray.
| Asset | Nominal CAGR | Real CAGR |
|---|---|---|
| Gold | 9.07% | 4.98% |
| US Equities | 10.84% | 6.68% |
And crucially, it is not gold's absolute return that justifies its presence in a portfolio. It is its correlation with other asset classes.
The numbers that change everything: near-zero correlation
Over the period January 1978 – February 2026, gold's monthly correlations are unambiguous:
- Gold vs. US equities: 0.05
- Gold vs. long-term bonds: 0.09
These figures mean that gold moves almost entirely independently of the two major traditional asset classes. This is not theory — it is what 48 years of real market data, post-Nixon, in the monetary regime we actually live in, consistently show.
An asset with a 0.05 correlation to equities does not amplify crises — it absorbs them. When equity markets collapse, gold does not fall with them. When inflation erodes bonds, gold holds. This is precisely the profile a retirement portfolio needs to navigate adverse return sequences — the sequence-of-returns risk that is the real enemy of a long retirement.
Is gold overvalued today? The question is better framed differently. The S&P 500-to-gold ratio measures how many ounces of gold it takes to buy the index — it places the relative valuation of both assets in long-term historical context. In May 2026, this ratio stands at approximately 1.57x. That is far from historical extremes in either direction:
| Period | S&P 500/Gold ratio | Context |
|---|---|---|
| Peak 2000 | ~5.5x | Tech bubble, gold unloved |
| Peak 1965–1968 | ~5.0x | Post-war economic boom |
| Trough 1980 | ~0.15x | Gold peak, stagflation |
| Trough 2011 | ~0.65x | Post-financial crisis |
| May 2026 | ~1.57x | Historical median zone |
Both equities and gold are expensive — but that is true of virtually every asset class in the current environment. The real question is not absolute valuation; it is portfolio balance.
What this means for a real-world retiree
SWR has a fundamental problem: it is designed for the worst historical case drawn from a dataset that does not form a coherent whole, and it imposes rigid withdrawals that adapt neither to market reality nor to the reality of a retiree's life. A bad market year does not mean you should withdraw the same amount as a good year. A retiree with variable expenses — which is everyone — does not need a blind autopilot.
A growing part of the FIRE community is questioning SWR in favor of adaptive approaches. The most rigorous, in my view, is VPW — Variable Percentage Withdrawal, developed by the Bogleheads community. Rather than fixing a withdrawal amount and hoping the portfolio survives, VPW calculates the optimal annual withdrawal based on the portfolio's actual current value, the remaining life horizon, and expected returns.
The result: withdrawals that automatically increase when markets are strong and decrease when they are difficult — exactly what any sensible wealth manager would do. VPW also automatically increases withdrawals as the time horizon shortens, allowing you to actually benefit from your capital while you still can, rather than involuntarily bequeathing it.
I have not yet published a dedicated English article on VPW, but the French version — VPW : la méthode de retrait adaptative que j'utilise en retraite — covers the mechanics and practical implementation in full. If the critique of SWR has convinced you there is a better way to run a retirement portfolio, that is the logical next step.
Frequently asked questions about the 4% rule
What is the 4% rule?
The 4% rule (also called the Safe Withdrawal Rate, or SWR) states that a retiree can withdraw 4% of their initial portfolio each year, adjusted for inflation, without depleting their capital over a 30-year retirement. It originates from the 1998 Trinity Study, based on historical US market data, and is widely cited as the benchmark withdrawal rate in the FIRE (Financial Independence, Retire Early) community.
Why is the 4% rule criticized?
The 4% rule relies on historical data spanning 150 years of radically different economic regimes — the gold standard, two world wars, 1970s hyperinflation, the post-Bretton Woods era. It imposes rigid withdrawals that do not adapt to real market conditions. In most historical scenarios, it leaves portfolios massively over-funded at the end of retirement — a sign of excessive conservatism that deprives retirees of usable capital. It is also poorly suited to the long retirements typical of early FIRE (40–50 years).
What is the alternative to the 4% rule?
The most rigorous alternative is VPW (Variable Percentage Withdrawal). Rather than a fixed amount, VPW calculates the optimal annual withdrawal based on the portfolio's current value, the remaining life horizon, and expected returns. Withdrawals rise in good market years and fall in bad ones, avoiding both premature depletion and involuntary accumulation.
Does gold belong in a FIRE retirement portfolio?
Yes — since 1971 and the end of Bretton Woods. Over the period January 1978 – February 2026, gold's monthly correlation with US equities is 0.05 and with long-term bonds 0.09 — near-total independence. In performance terms, over January 1972 – February 2026, gold delivers a real CAGR of 4.98% against 6.68% for equities. Before 1971, gold was an administered price fixed at $35/oz since 1934: its real CAGR was approximately -2.9% per year. Analyses that argue against gold in SWR backtests use precisely this pre-1971 data, contaminating the analysis with an asset that reflected no market behavior whatsoever. Since 1971, gold is a free-market asset and its near-zero correlation with both equities and bonds makes it an effective shock absorber against sequence-of-returns risk.
What is sequence-of-returns risk?
Sequence-of-returns risk is the danger that poor market returns occur early in retirement, when the portfolio is at its peak and withdrawals have their greatest impact. Unlike the accumulation phase — where market dips allow you to buy more cheaply — in the drawdown phase, downturns force you to sell more shares to maintain the same income. A portfolio sufficiently damaged early in retirement may never recover, even if markets subsequently rebound strongly.
Sources and data
- Jeremy Siegel, Stocks for the Long Run (6th ed., 2022) — long-run asset class performance data since 1801; updated to 2025 via WisdomTree — Jeremy Schwartz, January 2026
- Trinity Study (Philip L. Cooley, Carl M. Hubbard and Daniel T. Walz, 1998)
- Safe Withdrawal Rates: A Guide for Early Retirees (ERN)
- Richer Retirement Portfolio (portfoliocharts.com)
- Robert Shiller, Yale — historical US equity and inflation series since 1871: shillerdata.com
- Nominal and real CAGR gold/equities (Jan. 1972 – Feb. 2026): portfoliovisualizer.com
- Monthly correlations gold/equities/bonds (Jan. 1978 – Feb. 2026): portfoliovisualizer.com
- S&P 500 / gold ratio — long-term history: macrotrends.net
- Richer Retirement Portfolio — backtest and SWR analysis: dividendes.ch — Richer Retirement Portfolio by William Bengen
- VPW — adaptive withdrawal method: dividendes.ch — VPW (French)
- Variable Percentage Withdrawal (VPW) — Bogleheads Wiki
- Bengen, W. (2025). A Richer Retirement: Supercharging the 4% Rule to Spend More and Enjoy More.
En savoir plus sur dividendes
Subscribe to get the latest posts sent to your email.