Safe Withdrawal Rates in 2026: A Forward-Looking Simulation
The 4% rule grew out of historical withdrawal research, including William Bengen's 1994 work and the 1998 Trinity Study. Those studies asked how specified portfolios and spending rules would have performed through past return sequences.
This article asks a different question: what happens when we simulate future returns using the app's 2026 assumptions? It is a forward-looking model exercise, not a new historical Trinity backtest.
We generated 1,000 paths for each of three portfolio profiles and tested withdrawals over 15 to 40 years. The results depend on the return, inflation, fee and spending assumptions described below.
Four Percent Under These Assumptions
At a 95% modelled survival target over 30 years, the highest tested initial withdrawal rate for the balanced 60/30/10 portfolio is 3.0%. The conservative profile reaches 3.0%, and the aggressive profile reaches 2.5%. Rates were tested in 0.5 percentage point steps; these are grid results, not precise cut-offs or personal spending recommendations.
At a 4% initial withdrawal, 78.4% of the balanced portfolio's simulated paths survive 30 years. The remaining 21.6% deplete under these assumptions. That is a model result, not an estimate that the same proportion of real retirees will run out of money.
Run Your Own Numbers
Enter your portfolio allocation, pick a retirement horizon and modelled survival target, and get your safe withdrawal rate with a full survival heatmap, in seconds.
Open SWR CalculatorWhat We Tested
We tested three portfolio profiles across 13 withdrawal rates (2% to 8%) and 6 time horizons (15 to 40 years):
| Profile | Stocks | Bonds | Cash |
|---|---|---|---|
| Conservative | 30% | 60% | 10% |
| Balanced | 60% | 30% | 10% |
| Aggressive | 80% | 15% | 5% |
Simulation Parameters
Each simulation path models a retiree withdrawing a fixed percentage of their starting portfolio, adjusted annually for inflation, with annual rebalancing and a 0.5% management fee:
- Simulations: 1,000 paths per withdrawal rate/horizon combination
- Return model: Lognormal with Cornish-Fisher skew/kurtosis adjustment
- Inflation: Stochastic inflation using the USD parameters below; negative inflation is excluded
- Rebalancing: Annual
- Fees: 0.5% annual management fee
- Assumptions: J.P. Morgan 2026 return and volatility inputs, with Portfolio Lab correlations and simulation conventions
The model holds these assumptions constant throughout each 15-to-40-year horizon. J.P. Morgan's forecasts cover roughly 10 to 15 years; extending them beyond that window is an additional modelling assumption.
A Quick Recap: The Original Trinity Study
In 1998, three professors at Trinity University (Cooley, Hubbard, and Walz) published “Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable.” They tested withdrawal rates from 3% to 12% against rolling periods of U.S. stock and bond returns from 1926 to 1995.
For the inflation-adjusted 50/50 portfolio, the original study reported success in 95% of its historical 30-year periods at a 4% initial withdrawal rate. The underlying assets were the S&P 500 and long-term high-grade corporate bonds. Its 41 overlapping windows are historical observations, not independent forecasts of future retirees.
The historical results are useful evidence about sequence risk. Later research explores longer retirements, different markets, changing spending and additional income. These choices matter when comparing a result with the original study.
Karsten Jeske's Safe Withdrawal Rate Series examines many of these extensions. We link to his own explanation below rather than treat a withdrawal rate from a different setup as a direct validation of this model.
Why Forward-Looking Assumptions Matter
Historical backtests preserve the return sequences that actually occurred. They cannot establish which sequences will occur next. Forward assumptions let us test a different expected-return environment, but introduce forecast and model uncertainty of their own.
The model uses 7% nominal compound return for global equities and 4.8% for US aggregate bonds. Our comparison of 18 CMA providers shows that forecasts differ across firms. A different return, volatility or inflation assumption can change the withdrawal results.
The Results: Survival Heatmaps
Each cell shows the share of 1,000 paths that retain a positive portfolio balance at the horizon. Green meets the 95% target; yellow is 80% to below 95%, orange is 60% to below 80%, and red is below 60%. These colours describe the simulation, not a guarantee of safety.
Conservative Portfolio (30/60/10)
The bond-heavy portfolio performs best at lower withdrawal rates thanks to lower volatility, but it runs out of growth runway over longer horizons.
Simulated paths surviving (%), Conservative 30/60/10
| WR | 15y | 20y | 25y | 30y | 35y | 40y |
|---|---|---|---|---|---|---|
| 2.0% | 100.0% | 100.0% | 100.0% | 100.0% | 99.9% | 99.7% |
| 2.5% | 100.0% | 100.0% | 100.0% | 99.9% | 98.9% | 98.0% |
| 3.0% | 100.0% | 100.0% | 99.8% | 98.4% | 95.3% | 88.7% |
| 3.5% | 100.0% | 100.0% | 98.8% | 93.7% | 81.8% | 66.7% |
| 4.0% | 100.0% | 99.7% | 95.2% | 79.2% | 57.4% | 38.3% |
| 4.5% | 100.0% | 98.1% | 81.9% | 55.6% | 30.5% | 17.6% |
| 5.0% | 100.0% | 93.1% | 61.2% | 28.8% | 14.2% | 6.6% |
| 5.5% | 99.6% | 79.4% | 36.8% | 13.0% | 4.2% | 1.6% |
| 6.0% | 97.9% | 59.8% | 17.3% | 4.2% | 1.1% | 0.3% |
The 4% withdrawal rate has an 79.2% survival rate at 30 years and drops to just38.3% at 40 years. At the 95% survival target over 30 years, the highest tested rate is 3.0%.
Balanced Portfolio (60/30/10)
This profile holds 60% global equities, 30% US aggregate bonds and 10% cash. It differs from both a two-asset 60/40 portfolio and the original Trinity 50/50 example.
Simulated paths surviving (%), Balanced 60/30/10
| WR | 15y | 20y | 25y | 30y | 35y | 40y |
|---|---|---|---|---|---|---|
| 2.0% | 100.0% | 100.0% | 99.8% | 99.8% | 99.0% | 98.3% |
| 2.5% | 100.0% | 99.9% | 99.7% | 98.6% | 96.5% | 93.6% |
| 3.0% | 100.0% | 99.8% | 98.6% | 95.6% | 90.7% | 85.5% |
| 3.5% | 100.0% | 99.0% | 95.9% | 88.9% | 79.9% | 72.8% |
| 4.0% | 99.8% | 98.1% | 89.6% | 78.4% | 67.0% | 58.0% |
| 4.5% | 99.5% | 94.3% | 79.9% | 65.1% | 52.7% | 40.8% |
| 5.0% | 98.9% | 87.3% | 67.1% | 51.2% | 36.5% | 27.5% |
| 5.5% | 97.4% | 76.8% | 54.8% | 35.2% | 23.3% | 16.8% |
| 6.0% | 92.9% | 65.3% | 39.6% | 22.3% | 14.6% | 10.7% |
At 4%, the balanced portfolio survives 78.4% of simulated 30-year paths. The highest tested rate at the 95% survival target is 3.0%. At a 90% target, the grid result is 3.0%.
Aggressive Portfolio (80/15/5)
More equities means higher expected returns but also higher volatility. The sequence-of-returns risk is the killer here: a bad drawdown in the early years of retirement is catastrophic when you're simultaneously withdrawing.
Simulated paths surviving (%), Aggressive 80/15/5
| WR | 15y | 20y | 25y | 30y | 35y | 40y |
|---|---|---|---|---|---|---|
| 2.0% | 100.0% | 99.8% | 99.8% | 98.9% | 98.1% | 96.0% |
| 2.5% | 100.0% | 99.8% | 98.7% | 96.7% | 94.2% | 91.0% |
| 3.0% | 99.9% | 98.9% | 96.8% | 93.0% | 88.3% | 83.7% |
| 3.5% | 99.7% | 98.0% | 92.9% | 86.1% | 78.5% | 73.5% |
| 4.0% | 99.2% | 95.8% | 86.7% | 76.7% | 68.4% | 61.6% |
| 4.5% | 98.7% | 90.8% | 78.2% | 66.8% | 58.1% | 50.9% |
| 5.0% | 97.2% | 83.5% | 68.2% | 56.8% | 45.3% | 37.5% |
| 5.5% | 94.2% | 75.0% | 58.2% | 44.5% | 34.5% | 27.9% |
| 6.0% | 89.9% | 66.8% | 48.3% | 33.5% | 25.3% | 19.8% |
The aggressive portfolio performs slightly worse than the balanced portfolio at moderate withdrawal rates and horizons. At 4% over 30 years, survival is 76.7% versus the balanced portfolio's 78.4%. The highest tested rate at the 95% survival target is just 2.5%. (At very long horizons the extra growth finally pays: by 40 years the aggressive mix edges ahead.)
In this model, the extra equity return does not consistently offset the extra sequence risk at every horizon and withdrawal rate. That result depends on the assumptions and spending rule. Our Bitcoin retirement portfolio analysis explores another set of allocations.
Run Your Own Numbers
Enter your portfolio allocation, pick a retirement horizon and modelled survival target, and get your safe withdrawal rate with a full survival heatmap, in seconds.
Open SWR CalculatorSWR Summary: Highest Tested Rates Meeting the Target
| Profile | 20-year SWR | 25-year SWR | 30-year SWR | 35-year SWR | 40-year SWR |
|---|---|---|---|---|---|
| Conservative | 4.5% | 4.0% | 3.0% | 3.0% | 2.5% |
| Balanced | 4.0% | 3.5% | 3.0% | 2.5% | 2.0% |
| Aggressive | 4.0% | 3.0% | 2.5% | 2.0% | 2.0% |
All values are the highest tested rates meeting a 95% modelled survival target. At 30 years, each of these three profiles falls below 4% on this grid. The balanced 60/30/10 result is 3.0%.
How Does the Survival Target Change the Result?
The survival target is the fraction of simulated paths required to retain a positive balance through the horizon. A 95% target means at least 950 of 1,000 paths. It is not a statistical confidence interval or a guaranteed real-world probability. Both sampling variation and uncertainty in the assumptions affect the result.
| Modelled survival target | 30-year SWR | Annual from $1M |
|---|---|---|
| 99% | 2.0% | $20,000 |
| 95% | 3.0% | $30,000 |
| 90% | 3.0% | $30,000 |
| 85% | 3.5% | $35,000 |
| 80% | 3.5% | $35,000 |
At the 80% survival target, the balanced portfolio reaches 3.5% on this grid. The 4% rate does not meet any of the tested 80% to 99% targets in this particular model run.
How This Compares to ERN's Research
Karsten Jeske's Safe Withdrawal Rate Series uses historical returns and examines longer horizons, valuation-sensitive spending, glidepaths and supplemental cash flows. His toolbox lets readers change their own circumstances.
Our table uses simulated returns and one fixed inflation-adjusted spending rule. It does not test those dynamic strategies. Lower forward return assumptions are not the same as an explicit CAPE-based spending rule, and this exercise does not establish that the two methods agree on a suitable withdrawal rate.
What You Can Do About It
Use the table as a starting point for testing a plan. These variations change the question being modelled; their effects are not measured in the heatmaps above.
1. Flexible spending
A plan that reduces withdrawals after losses has a different spending path from this fixed real-spending baseline. Compare both the portfolio outcome and the spending cuts needed to achieve it.
2. Other income
Earnings or pension income can reduce the amount withdrawn from the portfolio while that income is received. Enter its amount and timing explicitly; a temporary income stream is not equivalent to a permanent increase in the withdrawal rate.
3. Benefit timing
Pension and Social Security start dates affect the cash flows a portfolio must fund. This experiment includes neither, so its result does not compare claiming strategies.
4. Changing the allocation
A glidepath changes asset weights during retirement. The portfolios above rebalance to constant weights, so a glidepath needs a separate test with an explicit schedule.
5. Fees
The model charges 0.5% annually. Lower fees leave more money invested under the same return path. The change in a sustainable withdrawal rate still needs to be calculated; it is not necessarily equal to the fee reduction.
Important Caveats
Forward-looking assumptions aren't predictions
J.P. Morgan's capital market assumptions describe expected returns over 10 to 15 years. They are inputs to this model, not knowledge of future outcomes. Holding them fixed for up to 40 years extends beyond their stated forecast horizon.
Monte Carlo has limitations
The engine samples annual returns with a lognormal transformation and Cornish-Fisher skew and kurtosis adjustments. Years are drawn independently under constant parameters. It does not reproduce persistent market regimes or a specified historical crisis. Its inflation sampler also excludes deflation and caps annual inflation at 10%. These choices limit what the survival frequencies can establish.
The 0.5% fee assumption matters
If you use low-cost index funds with expense ratios around 0.05-0.10% and no advisor fee, your SWR would be higher. Conversely, if you pay a 1% advisor fee plus fund expenses, it would be lower.
This is a fixed-withdrawal analysis
This baseline keeps the initial spending amount constant in inflation-adjusted terms. It excludes spending adjustments, taxes and additional income. A rate that meets the modelled survival target can still fail in some paths; it is not a guaranteed floor.
For the history of the 4% rule, see AAII's review of the withdrawal-rate studies, which distinguishes Bengen's 1994 work from the later Trinity research.
Further Reading
- Big ERN's Safe Withdrawal Rate Series: A collection of historical withdrawal studies and tools. Start with the series guide.
- The original Trinity Study: Cooley, Hubbard, and Walz (1998),Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable.
- J.P. Morgan 2026 LTCMA: Published annually by J.P. Morgan Asset Management. See the LTCMA publication; this model uses the 2026 edition.
- Wade Pfau's research: See How Much Can I Spend in Retirement? for an accessible overview of modern SWR research.
Go Deeper With Portfolio Lab
Full Monte Carlo withdrawal analysis is just one module. Portfolio Lab also optimizes across 57 asset classes, backtests back to 1926, and runs simulations of up to 10,000 paths, on the published forecasts of J.P. Morgan, BlackRock, Research Affiliates and AQR, averaged and marked to today's prices.
Start Optimizing, FreeNo credit card required
Full Methodology
All results were computed using Portfolio Lab's withdrawal analysis engine. The simulation generates 1,000 Monte Carlo paths with the following parameters:
- Starting value: $1,000,000
- Return model: Lognormal with Cornish-Fisher skew/kurtosis adjustment
- Asset inputs: USD global equities (ACWI), US aggregate bonds and cash; nominal compound return and volatility inputs from the app's 2026 J.P. Morgan table
- Correlation structure: Portfolio Lab's historical correlation estimates with Stambaugh adjustment and positive-semidefinite repair, rather than a J.P. Morgan correlation matrix
- Inflation: Stochastic USD input, 2.57% mean parameter and 1.77% volatility parameter; mean from the Cleveland Fed ten-year model (2026-09), volatility retained from J.P. Morgan 2026; annual draws clipped to 0% to 10%. The sampler mixes the weighted asset shock with an independent shock using a 0.3 coefficient; realised correlation with portfolio returns need not equal 30%.
- Withdrawals: Fixed percentage of starting value, inflation-adjusted annually
- Rebalancing: Annual to target weights
- Management fee: 0.5% annually
- Withdrawal rates tested: 2.0% to 8.0% in 0.5% steps
- Horizons tested: 15, 20, 25, 30, 35, 40 years
- SWR defined as: highest tested withdrawal rate with >= X% survival (X = the modelled survival target)
- Seed: 42 (deterministic for reproducibility)
The article figures are regenerated from the same simulation engine and configuration as the calculator and checked for agreement. Use the profiles, fee, inflation mode and seed above in the SWR calculator.