Is the 4% Safe Withdrawal Rate Really Safe? - The 2.7% Rule for Retirement Spending 

You have probably heard of the four percent rule for retirement spending which says that you can safely spend four percent of an Investment Portfolio in the first year of retirement and then adjust that dollar amount for inflation each year for the rest of your life with minimal risk of running out of money. The four percent rule relies on biased data. Recent research corrects for these data biases and suggests a safe withdrawal rate that is somewhere between two and three percent depending on the portfolio being tested and the life expectancy of the investor.

2.7 (let's call it 3) percent is the new four percent for safe retirement spending. Financial planner William Bengan wrote a paper in 1994 determining withdrawal rates using historical data. Bengan took historical data for U.S stocks and intermediate term treasuries and tested how long a portfolio of 50 stocks and 50 bonds would be able to sustain various levels of withdrawals stated as a starting percentage of the portfolio and adjusted for inflation. Thereafter the result is an historically safe spending rate which Bengan determined to be four percent in his data.

In the 2022 paper 'popular personal financial advice versus the professors' James Choi reviews the 50 most popular personal finance books and finds that most of them recommend a four percent or higher safe spending rule. According to a recent survey from Vanguard 22 percent of Millennials are planning for early retirement based on the four percent rule. Anecdotally I can tell you that lots of people holding themselves out as personal finance experts do indeed advocate for using the four percent rule. The primary problems with the four percent rule are that it was based on a 30-year withdrawal period which may not be long enough for many people today and it was based on U.S stocks and bonds. 

We know today, looking backward, that the U.S has been one of the best performing Equity markets in the world but it's less clear that the U.S experience is representative of expected returns going forward informing expectations. It's important to ask why we got the outcome that we did before assuming that it's going to repeat itself. U.S stock returns have been so much higher than economic models would have predicted them to be that the phenomenon has been referred to as the equity premium puzzle There have been many events historically that did not happen which if they had happened would likely have materially affected the experience of U.S investors. 

For example, the U.S was not heavily impacted economically on their home soil by either World War and the Cuban Missile Crisis was resolved peacefully without the economic devastation that could have occurred in an alternate reality. U.S investors were compensated for taking the risk that these catastrophic outcomes could have materialized and they were compensated for the good fortune that none of them did to make this point quantitatively.

The 2022 paper is "the United States a lucky Survivor - a hierarchical Bayesian approach" evaluates the effects of survivorship and luck on realized U.S Equity returns from the perspective of an investor in 1920. They find that their realized historical risk premium on U.S stocks, exceeds the expected premium by two percent. This excess return is approximately equally split between contributions from luck where cash flows ended up being higher than expected due to disasters that did not materialize and learning where investors lower their required return on U.S stocks over time as catastrophes do not happen driving up U.S equity valuations.

Looking Backward at the return on U.S stocks and assuming that they will repeat is not only a bet that there will be more good luck but that there will be enough good luck to more than offset the currently low expected returns. An alternative approach, and the one that I think is much more sensible in developing a safe withdrawal rate is to draw from the full sample of developed Equity markets including the ones that had bad returns or failed completely and that is exactly what is done in the 2022 paper on the safe withdrawal rate evidence from a broad sample of developed markets.

The authors use a comprehensive data set of real returns for domestic Equity, International equity and government bonds in developed economies to investigate safe withdrawal rates. The data covers approximately 2500 years of asset class returns in 38 developed countries over the period 1890 to 2019. They include data for countries that don't typically make it into historical data sets because the market failed or the data are otherwise difficult to obtain using this survivorship and easy data bias.

Corrected data set the authors test save withdrawal rates they test a portfolio of 60 domestic stocks and 40 domestic bonds. Various alternative asset allocations from zero percent domestic stocks all the way up to a hundred percent, a Target date fund that shifts more into bonds over time and in unreported analysis one of the co-authors also, tests 60 stock and 40 Bond portfolios with various levels of international stock diversification they use a block bootstrap simulation method to draw samples from the historical data the result is a broad range of possible investment experiences that draws on the range of possible outcomes across the countries and time periods in the sample rather than using a fixed 30-year withdrawal period like the original four percent rule research did they use mortality tables from the U.S Social Security Administration to incorporate longevity risk into their analysis. 

Based on this data the mean life expectancy for a couple age 65 in 2022 is 24.7 years but the fifth percentile is 12.3 years whereas the 95th percentile is 35.5 years they include this variability in their simulations with this setup they find for a 65 year old couple in 2022 investing in domestic developed market stocks and bonds that the four percent rule has a 17.4 percent chance of depleting Financial wealth prior to death and a 16 percent chance of depleting wealth and living another five years allowing for a five percent probability of financial ruin. 

If they fund a 2.26 percent safe withdrawal rate for a domestic investor which is clearly much less than four percent across alternative asset allocations from zero percent stocks through 100 stocks domestic stocks they find that the 60 40 portfolio gives the highest safe withdrawal rate so at least in this data getting more aggressive with stocks doesn't help to increase safe spending they also find that the target date fund underperforms the 60 40 portfolio adding an allocation to International stocks to the 60 40 portfolio improves the numbers for that same 2022 retiree.

Moving the portfolio 40 percent of the equity portion of the portfolio into International stocks improves the safe withdrawal rate to 2.85 percent and moving 90 percent of the equity portfolio into International stocks and brings it to 3.02 percent these numbers account for an additional 0.5 percent in costs for owning International stocks over domestic stocks to try and capture the typically higher fees and less favorable tax treatment in both taxable and non-taxable accounts.

The 2085 American retiree can safely sustain a spending rate closer to 2.7 percent with an internationally Diversified 60 40 portfolio. These figures do not account for taxes so a taxable investor may need to revise the numbers down further.

There are a couple of important lessons here. One is that the four percent rule is not a sustainable spending rate when you look outside the U.S and account for the risk of living a long life. The other is that international diversification matters a lot.

The good news is that withdrawal rates are not that useful anyway. We don't use them in financial planning. The constant withdrawals used in safe withdrawal rate analysis are inferior to variable withdrawals for sustainable spending. Don't blindly follow a withdrawal rule all the way down to a depleted portfolio. While they're still alive people will make adjustments to their spending or they'll find ways to make income. 

On top of that the required portfolio withdrawals needed to maintain spending will likely decrease over time as things like government pensions kick in and those government pensions will help to hedge against the risk of living longer than expected.

There are other strategies too like deferring government pensions as long as possible to increase their benefit or allocating a portion of the retirement portfolio to annuities 

Beyond that some empirical evidence suggests that retirees don't increase their spending with inflation over time they tend to increase their spending at a rate that TRAILS inflation by about one percent. Safe withdrawal rates typically assume constant inflation adjusted withdrawals.

Finally, the data we have discussed are based on investing in a market capitalization weighted index portfolio of domestic and international stocks and government bonds tilting a portfolio towards smaller and lower priced stocks and corporate bonds which theory and evidence suggest have higher expected returns, is likely to help improve safe withdrawal rates but importantly we're talking about improving a baseline 2.7 percent safe spending rate not a four percent rate.

Most safe withdrawal rate analysis supporting a four percent safe spending rate is based on historical U.S data which one of the best performing financial markets in documented history.

Drawing on a comprehensive data set that accounts for survivorship and easy data biases spanning 38 developed countries from 1890 to 2019 and accounting for longevity risk gives a much more realistic and sobering view on the safe withdrawal rate which I estimate at 2.7 percent.