How do safe withdrawal rates vary with starting valuations?

The short answer

A safe withdrawal rate is not a constant of nature. It depends heavily on the price you paid for assets at retirement entry. The same 4% rule that comfortably worked for a 1982 retiree (low CAPE, high real bond yields) failed or nearly failed for a 1966 retiree (elevated CAPE, subsequent stagflation).

The cleanest empirical evidence comes from Michael Kitces’s 2008 analysis (extended through 2024) showing that Shiller CAPE at the start of retirement has a strong inverse relationship with the maximum 30-year SWR. The relationship is mechanical: high CAPE implies lower forward equity returns, which compounds badly with the sequence-risk dynamics of early retirement.

The practical implication is that the same nominal withdrawal rule embedded in a financial plan should be calibrated against current valuation conditions. A 4% rule starting at low CAPE is conservative; the same rule starting at very high CAPE is aggressive.

→ New to valuation-aware withdrawal frameworks? Financial education hub

What the data shows

The reference numbers come from Kitces (2008, updated 2024), Bogleheads.org research extending the dataset through 2017, and Shiller’s online data plus current market figures.

The reference numbers (Kitces 2008-2024 and Shiller data):

• R² between starting Shiller CAPE and 30-year SWR — approximately 0.77 for portfolios with at least 60% equity allocation
• Shiller CAPE current value (May 2026) — 40.11, near record highs
• Shiller CAPE historical median — 16.05, with record low of 4.78 in the early 1920s and record high of 44.2 during the dot-com bubble
• Bottom CAPE quintile (low valuations) — historical SWRs typically 6%+ for 30-year retirements with 60/40 portfolio
• Top CAPE quintile (high valuations) — historical SWRs typically 3-4% for 30-year retirements with 60/40 portfolio

The exception that complicates the picture: CAPE is a noisy predictor over short horizons (1-5 years) and only becomes statistically meaningful over 10-20 year periods. This horizon happens to coincide exactly with the sequence-risk window of retirement, which is why the CAPE-SWR relationship is empirically strong despite CAPE being a poor short-term timing tool.

→ Dataset: Real interest rates vs CAPE ratio dataset

Why it happens — the macro mechanism

The valuation-SWR link is the joint product of mean reversion in equity returns and the asymmetric arithmetic of withdrawals. Starting valuations contain information about the next 10-20 years of returns, and those returns dominate the sequence-risk window.

The first channel is forward equity return predictability. Shiller demonstrated using more than 130 years of US data that high CAPE periods correlate with poor 10-20 year forward equity returns. The mechanism is mean reversion in long-run earnings yields: when prices are high relative to smoothed earnings, future returns must come predominantly from earnings growth and dividends rather than multiple expansion.

The second channel is the alignment between CAPE-prediction-horizon and sequence-risk window. Kitces noted that CAPE predicts 10-15 year returns reasonably well, and this is exactly the window during which sequence-risk dynamics dominate the 30-year SWR. A retiree facing weak first-15-year returns (because of high starting CAPE) gets the worst possible sequence: lower returns combined with active withdrawals.

A brief transition: this does not mean CAPE is a market timing tool. It is a base-rate input that informs the prudent SWR distribution, not a buy/sell signal.

The third channel is the bond return interaction. Real bond yields at retirement entry act as a complementary input. High CAPE plus low real bond yields (as in 2020-2021) is the worst environment for SWR. High CAPE plus higher real bond yields (where current 2026 conditions partly sit) provides some offset because the bond portion of a balanced portfolio earns more.

Synthesis by regime: in the bottom CAPE quintile combined with high real bond yields (1982 retirees), SWRs of 6-8% were sustainable. In the median CAPE quintile (1990s pre-bubble retirees), 4-5% was typical. In the top CAPE quintile combined with low real bond yields (2000 retirees, partly 2021-2022 retirees), SWRs of 3-4% were the historical pattern. The transition parameter is the joint state of CAPE quintile and real bond yield level at retirement entry, which together explain most of the historical SWR variance.

The price you pay for assets at retirement entry quietly determines what you can spend for the next thirty years.

→ Framework: Asset allocation and resilient portfolios

What it means for different economic actors

Pre-retirees entering a high-CAPE environment may want to plan for the lower part of the historical SWR distribution rather than the average, accepting more conservative initial withdrawals as a buffer against bad sequence outcomes.

Retirees already in withdrawal can recalibrate based on realized first-decade returns, since the worst sequence-risk damage is partly observable rather than purely forward-looking.

Pension funds and endowments with multi-cohort horizons can average out sequence risk across rolling vintages, but this option is not available to individual retirees.

A common error is to use a single fixed withdrawal rate without periodic recalibration based on portfolio performance and current valuation conditions. Dynamic frameworks (Guyton-Klinger guardrails, Kitces ratcheting rules) explicitly address this by adjusting withdrawals as conditions evolve.

Practical observation

Go deeper

Related questions

• Why does the 4% rule face challenges in current markets?
• What is sequence of returns risk?
• Is the CAPE ratio reliable for valuations?
• What is longevity risk?

Frequently asked questions

Is the CAPE-SWR relationship reliable for non-US markets?

The empirical work by Kitces and others is overwhelmingly based on US data, where 100+ years of consistent series exist. International evidence is weaker but generally consistent: starting valuations have predictive power for forward 10-15 year returns in most developed markets. For retirees diversified globally, the relevant CAPE input is a weighted blend of regional valuations rather than US CAPE alone.

How sensitive is the SWR to bond yield assumptions?

Very sensitive. Pfau, Finke, and Blanchett demonstrated that under -1.4% real bond return assumptions, even a 4% withdrawal rule failed in 57% of Monte Carlo simulations. Conversely, real bond yields above 2% historically supported SWRs comfortably above 5%. The bond yield assumption is at least as important as the equity valuation assumption, and often more so for balanced portfolios.

Should the SWR be adjusted dynamically over retirement?

The literature increasingly favors dynamic frameworks over fixed-rate approaches. Guyton-Klinger guardrails set initial withdrawals slightly higher than worst-case SWR but adjust downward when portfolio performance deteriorates. Kitces ratcheting rules raise withdrawals when portfolios perform well. These approaches respect the empirical reality that fixed-rate planning either over-saves in benign environments or runs out of money in bad sequences. The trade-off is added complexity and the discipline to actually implement the adjustments.

Last updated — 4 June 2026