How much can retirees safely withdraw from their portfolios?

That’s one of the issues we’ve been addressing in our annual The State of Retirement Income report since November 2021. The study differs from some other widely used research on safe withdrawal rates in that it relies on forward-looking estimates for future market returns, as opposed to testing the numbers based on historical data. William Bengen’s landmark research in 1994, for example, found that a static withdrawal rate of 4% of the initial portfolio balance with the dollar amount adjusted each year for inflation would have never failed over any 30-year period going back to 1926.

But because future market returns may or may not be in line with the past, we incorporate forward-looking return and inflation estimates from our colleagues at Morningstar Investment Management to estimate how much retirees can safely spend going forward.

For each hypothetical withdrawal percentage, we calculate the probability of success based on simulated results for 1,000 potential return scenarios over a 30-year period. The goal is to find the highest withdrawal percentage that ends up with a positive portfolio balance in at least 90% of these return paths. We test a variety of flexible spending methods, along with a “base case” that assumes retirees keep annual withdrawals constant in dollar terms by adjusting each year’s withdrawal amount for inflation.

The resulting recommended withdrawal rate is meant to be a general guideline, not a strict edict. People willing to use a more flexible approach or a lower probability of success can safely withdraw larger amounts (and in fact, a more flexible approach to withdrawals would probably be the best practice for the majority of retirees). But for new retirees who followed the suggested baseline amount for retirement withdrawals starting in 2022, our initial research has worked out well so far.

The results

In the first report published in 2021, my colleagues Christine Benz, Jeff Ptak, and John Rekenthaler (who has since retired) recommended that new retirees take a conservative approach to retirement income by using a starting withdrawal rate of 3.3%. The 3.3% figure was based on cautious estimates for future asset-class returns, which increase the likelihood that portfolio balances will run out before the end of an assumed 30-year retirement period.

The report’s 3.3% recommendation raised a few eyebrows. If a 4% withdrawal rate has always worked historically, people wondered, why use a number that’s so much lower? Assuming a $1 million starting portfolio balance, retirees withdrawing 3.3% per year would end up with a first-year withdrawal amount of just $33,000, versus $40,000 based on a 4.0% withdrawal rate. Depending on a retiree’s other sources of income, that difference could mean a lot of belt-tightening or forgoing some travel plans during the first year of retirement—exactly when it might be nice to spend a little more lavishly.

But taking a more cautious approach to retirement spending turned out to be a wise move for people who started retirement at the beginning of 2022. By the end of the year, both stocks and bonds had suffered double-digit losses, leading to a 15% loss for the 50/50 portfolio I tested for this article. (The portfolio I tested is based on a 40% weighting in US stocks, 10% in international stocks, 32% in US core bonds, 8% in global core bonds, and 10% in cash, rebalanced annually.) Inflation would have been a double whammy, as it rose to about 6.5% in 2022 and remained above average in the following years. That means retirees would have had to withdraw more each year just to keep inflation-adjusted spending flat, which also cut into portfolio values.

For retirees who adopted an initial withdrawal rate of 3.3%, the portfolio would have dropped as low as about $786,000 after withdrawals made at the beginning of 2023 but bounced back to just over $1 million by the end of 2025. At the same time, the inflation-adjusted withdrawal amount would have risen to about $38,400 by the beginning of 2026, which is 3.66% of the balance.

I ran these numbers through the same Monte Carlo simulation used in our other research (using the same capital market assumptions we used in the most recent edition of the paper) and found that 966 out of 1,000 trials (96.6%) would have ended up with a positive balance over the remaining 26 years in the test period.

That’s a pretty high probability of success and safely clears the 90% hurdle we set for our initial recommendation. In fact, these odds are high enough that retirees who followed our conservative approach starting in 2022 can probably afford to bump up their annual spending to about $40,800 and still clear the 90% hurdle, based on our most recent return estimates.

Retirees who followed the standard 4% rule would have enjoyed less favorable results thus far, as a more aggressive approach to spending exacerbated the impact of poor market performance plus above-average inflation. The portfolio value would have dropped as low as about $773,000 by the beginning of 2023 but bounced back to slightly above $1 million by the end of 2025.

That’s only about $38,000 lower than the year-end balance for a retiree who started out with a 3.3% withdrawal rate. But because of the ripple effects of inflation on annual spending amounts, retirees who started out with a $40,000 withdrawal amount would now be spending more than $46,000 per year—a withdrawal rate of 4.6% based on the portfolio balance at the end of 2025.

The probability of success over the next 26 years looks significantly lower based on these numbers, dropping to 85.2%. Some retirees might be fine with those odds, but people who prefer a greater level of certainty may want to pull back slightly on spending. Cutting annual spending (ideally to about $39,400) would improve the odds of success.

Final thoughts

While the initial results for retirees following our recommendations have been positive, it’s worth noting that we won’t get it right every year. Estimating future market returns is notoriously difficult, particularly over a 30-year period. Inflation is another major question mark. The model used in our most recent paper assumes a 2.42% inflation rate, but if actual inflation turns out to be higher going forward, retirement spending will be less sustainable. Even so, the initial results from our first foray into withdrawal rate research underscore the value of a conservative approach.