This article is part of a regular series of thought leadership pieces from some of the more influential ETF strategists in the money management industry. Today's article is by Craig Israelsen, Ph.D., creator of the 7Twelve portfolio, consultant to 7Twelve Advisors, LLC and executive-in-residence in the Financial Planning Program at Utah Valley University.
What is a prudent withdrawal rate from a retirement portfolio?
Retirees necessarily face this complex question as they attempt to balance the need for retirement income (which may push them toward a higher withdrawal rate) and the fear of exhausting their retirement portfolio prematurely (pushing them toward a lower withdrawal rate).
The analysis presented here evaluates a wide range of retirement portfolio withdrawal rates, and is based on a retirement portfolio comprising four primary asset classes: large U.S. stock, small U.S. stock, U.S. aggregate bonds and U.S. cash. A more broadly diversified portfolio would have been preferred, but only these four asset classes (based on the performance of underlying indexes) have reliable performance data back to 1961.
The performance of large cap U.S. stock was represented by the S&P 500 Index, while small U.S. stock was represented by the Ibbotson Small Stock index from 1961 to 1978 and the Russell 2000 Index from 1979 to 2019. U.S. bonds were represented by the SBBI U.S. Intermediate Government Bond Index from 1961 to 1975, and the Barclays Aggregate Bond Index from 1976 to 2019. U.S. cash was represented by 90-day Treasury bills.
The portfolio asset allocation in this analysis was 40% large U.S. stock, 20% small U.S. stock, 30% bonds and 10% cash (with annual rebalancing). A total portfolio cost of 100 basis points was assumed (100 bps was subtracted from the annual returns of the index-based retirement portfolio).
Studying the results of retirement portfolio survival during only one 25-year period is inadequate given the variability of the returns of the investment portfolio. Thus, multiple 25-year time periods need to be analyzed, which is exactly what was done in this analysis.
In fact, 35 rolling 25-year periods were examined between 1961 and 2019. The first 25-year retirement period was from the start of 1961 to the end of 1985. The second 25-year period in which money was withdrawn annually from a retirement portfolio was from 1962 to 1986, and so on. The 25-year period represents, for example, a retiree between the ages of 65-90, or 70-95 or 75-100, etc.
Start With $1 Million
A starting balance of $1 million was assumed at the beginning of retirement. The four-asset retirement portfolio was tested using 15 different withdrawal rates ranging from 1% to 15% for each of the 35 rolling 25-year periods. A 1% withdrawal rate indicates that 1% of the portfolio’s balance at year end was withdrawn by the retiree.
This analysis did not use a cost of living adjustment increase, commonly known as a COLA. The results were compiled and averages computed. The impact of taxes was not accounted for. This analysis also assumes the retirement portfolio was not subject to the required minimum distribution (RMD). An example of a retirement account that is exempt from RMD requirements is a Roth IRA.
As shown in Figure 1, a 1% annual withdrawal rate produced an average annual withdrawal of $33,962. Recall that this figure represents the average over 35 rolling 25-year periods between 1961 to 2019. The average ending portfolio balance (after 25 years of annual withdrawals) was $8.119 million.
Clearly, a four-asset portfolio can handle a 1% withdrawal rate inasmuch as the average ending balance was over 8x larger than the starting balance. Assuming a 1% of portfolio balance annual withdrawal rate, the maximum ending balance was nearly $19 million, while the smallest ending balance was just under $4 million. Thus, a 1% withdrawal rate is clearly conservative.
4% Withdrawal Rate
Let’s consider the well-known 4% withdrawal rate. The average annual withdrawal was $88,182, and the average ending portfolio balance 25 years later was just over $4 million. The maximum ending balance was $9.49 million, and the smallest ending balance was $1.8 million.
The variance between the maximum and minimum ending balance clearly reveals the impact of what we refer to as “sequence-of-returns risk.” If the sequence of portfolio returns was favorable (meaning high returns in the early years) the ending balance is much higher, and vice versa.
It’s important to note that at every withdrawal rate (from 1% to 15%), the portfolio survived intact for all 25 years in all 35 rolling 25-year periods. This is the result of withdrawing only a set percentage (from 1% to 15%) of the portfolio’s ending balance each year rather than annually withdrawing an escalating amount from year to year over the 25-year periods.
Withdrawing a percentage amount allows the annual withdrawal to decrease in those years that the portfolio suffered a market-based loss. This is a self-preserving retirement portfolio withdrawal technique.
As shown in Figure 1, the amount of the annual withdrawal increased each year 68% of the time at a 4% annual withdrawal rate. Or, conversely, the annual withdrawal decreased the following year 32% of the time over the rolling 25-year periods.
Interestingly, the average decrease in annual withdrawal in those years in which they decreased was -$6,064—a comparatively small decline compared to the average annual withdrawal of $88,182 (at a 4% withdrawal rate).
When only withdrawing a percentage of the portfolio’s value at the end of each year, it’s not possible to drive the portfolio balance to zero. Practically speaking, however, the value of the portfolio will be seriously eroded at high rates of withdrawal (as shown by the minimum ending balance using withdrawal rates in excess of 10%).
6% Withdrawal Rate
A 6% withdrawal rate is highlighted in yellow because it represents the highest rate that never produced an ending balance lower than the starting balance of $1 million. A withdrawal rate of 8% is highlighted in green because it represents the rate that maximized the average annual withdrawal.
Figure 1: Withdrawal Rate Comparison
Results derived from the analysis of 35 rolling 25-year retirement periods from 1961-2019
$1 million assumed starting balance in retirement account
Asset allocation: 40% large U.S. stock, 20% small U.S. stock, 30% bonds, 10% cash
100 bps annual portfolio cost subtracted from index-based portfolio returns
Raw data source: Steele Mutual Fund Expert, calculations by author
It may be counterintuitive, but once the withdrawal rate exceeded 8% in this analysis (using a 40/20/30/10 asset allocation) the average annual withdrawal began to decline. In other words, employing a withdrawal rate higher than 8% did not increase retirement income.
Said more plainly, we should not use a withdrawal rate higher than 8% if our retirement portfolio has an allocation similar to the one in this analysis, as it will not likely increase our average annual withdrawal.
Furthermore, an annual withdrawal rate of 10% or higher caused the average ending balance of the retirement portfolio to be lower than the starting balance of $1 million.
The interplay between withdrawal rate, average annual withdrawal amount and ending portfolio balance after 25 years is depicted in Figure 2. The average ending balance of the retirement portfolio over 35 rolling 25-year periods is shown by the size of the blue dots—which uniformly decrease in size at each higher withdrawal rate.
The location of the blue dot on the y-axis depicts the size of the average annual withdrawal, which systematically increases up to an 8% withdrawal rate and then begins to decline after that point.
Figure 2: Interaction of Withdrawal Rate, Ending Balance and Average Annual Withdrawal
Analysis of 35 rolling 25-year retirement periods from 1961-2019
Starting balance of $1 million
40% large stock, 20% small stock, 30% bonds, 10% cash
100 bps assumed portfolio cost subtracted from index-based portfolio returns
In conclusion, a prudent annual withdrawal rate from a retirement portfolio will range from 3% to 6%. Retirees using a withdrawal rate closer to 6% will understandably be more susceptible to depleting their portfolio more rapidly.
However, over the 35 rolling 25-year periods from 1961-2019, a retirement portfolio experiencing a 6% annual withdrawal rate never had an ending balance after 25 years that was lower than the starting balance. Moreover, a 6% annual withdrawal rate allowed the retiree to experience a larger annual withdrawal year-over-year 65% of the time.
A secondary, but perhaps more important, observation from this analysis is that the retirement portfolio survived intact for the full 25 years in every rolling period—regardless of the size of the withdrawal rate. Furthermore, at withdrawal rates of 9% or lower, the average ending balance 25 years later was larger than the starting balance.
At the very least, this analysis should allow retirees to live with less fear that their portfolio will die before they do. Assuming a reasonably diversified portfolio and an annual withdrawal rate of 6% or less of the portfolio’s balance, it is highly likely the portfolio will outlive the retiree.
These results, as already mentioned, are based upon a four-asset class retirement portfolio with an assumed cost of 100 bps. As shown in my article last month, a broadly diversified ETF-based portfolio can be built for 10 bps.
The impact of a 90 bps cost reduction in a retirement portfolio is significant. For example, at a 4% withdrawal rate and 100 bps of portfolio cost, we saw that the average annual withdrawal was $88,182 over the 25-year retirement period. By lowering the portfolio cost to 10 bps, the average annual withdrawal increased to $100,146—an improvement of $11,964. This translated to $997 more retirement income per month!
The spreadsheet tool used to perform the analysis reported in this article is available for purchase from the author. His email is [email protected].