Last year Morningstar told us that the 4% Rule was really 3.3%. This year, based on revised assumptions about the future, they tell us that it is now 3.8% (The State of Retirement Income 2022). Of course, this new report has set off another flurry of discussions among financial advisors, financial academics and the retirement-focused media about:
- the 4% (or whatever %) Rule,
- Safe Withdrawal Rates and
- Spending under more-flexible spending strategies (dynamic vs. static spending strategies)
In this post, we will
- Discuss the results of the Morningstar 2022 Report,
- Briefly discuss the difference between static approaches like the 4% Rule and static Monte Carlo model projections and dynamic approaches like the Actuarial Financial Planner or dynamic Monte Carlo projections, and
- compare withdrawal rates developed in the 2022 Morningstar report with initial withdrawal rates developed from a simplified application of our Actuarial Financial Planner (AFP).
We conclude, once again, that the AFP is a much more robust planning tool than the 4% Rule or most Monte Carlo models.
Morningstar Report 2022
According to the Morningstar report authors, Christine Benz and John Rekenthaler:
“Because equity valuations have declined and cash and bond yields have increased, the forward-looking prospects for portfolios—and in turn the amounts that new retirees can safely [emphasis added] withdraw from those portfolios over a 30-year horizon—have enjoyed a nice lift since we explored the topic last year. Whereas last year’s research suggested that a 3.3% withdrawal rate was a safe starting point for new retirees with balanced portfolios over a 30-year horizon, this year’s research points to 3.8% as a safe starting withdrawal percentage, with annual inflation adjustments to those withdrawals thereafter.”
The authors also note:
“Retirees who are willing to employ more-flexible [or dynamic] strategies or make other modifications to a fixed real withdrawal system can enjoy even higher starting withdrawals, assuming they’re willing to accept other trade-offs, such as fluctuating year-to-year real cash flows and the possibility of fewer leftover assets at the end of a 30-year period.”
The authors define “safe” for the purposes of determining a safe withdrawal rate from a portfolio as one having a 90% probability of not running out of money over a 30-year time horizon based on the assumptions discussed below. This probability may also be viewed as a 10% probability of having to reduce future spending. Under this static withdrawal approach, an initial withdrawal amount is established and increased in subsequent years by increases in CPI each year, without adjustment up or down to reflect actual investment experience. In addition, the retiree is assumed to spend exactly this “safe” amount (no more and no less) each year.
This year’s report used the following assumptions (which the Morningstar authors claimed to be “conservative”) to develop the 3.8% safe withdrawal rate for a portfolio invested 50% in stocks and 50% in bonds:
- Mean annual stock investment return: 9% to 12% depending on type of stock with US equities at the lower end of the range. For our comparisons below using Morningstar assumptions, we assumed 9.5%. By comparison the current default assumption for investment return on risky assets under the AFP model is 7.5%
- Mean annual bond returns: 5% per annum. By comparison, the AFP default assumption for investment return on non-risky assets/investments is 4.5%
- Mean inflation: 2.8%. By comparison, the AFP default assumption for inflation is 3.5%
- Lifetime planning period: 30 years. By comparison, the AFP default assumption for lifetime planning period is the 25% probability of survival for non-smoking males or females in excellent health based on the Actuaries Longevity Illustrator. For a 65-year old male, the lifetime planning period is 29 years.
In addition to the safe withdrawal rate of 3.8%, the 2022 Morningstar report contained the following interesting conclusions with respect to the impact of varying investment allocations on safe withdrawal rates:
“Over normal retirement time horizons of 25 to 30 years, balanced asset allocations support the highest starting withdrawal amounts—higher than equity-heavy allocations. In fact, an investor could dial the portfolio’s equity allocation all the way down to 30% of assets, with the remainder in fixed income and cash, employ a 3.8% starting withdrawal with annual inflation adjustments thereafter, and still have a 90% chance of not outliving the money over a 30-year period. (However, as discussed later, taking the more conservative path would likely reduce the portfolio’s final value when the 30-year period concluded.) Over shorter time horizons—10 and 15 years, for example—bond-heavy allocations support higher starting withdrawals than do equity-heavy ones. Bonds’ now-stronger return expectations (thanks to their higher yields), along with their lower volatility relative to equities, are the key reason that balanced to-conservative asset allocations support higher withdrawal rates.”
The authors note that since spending research appears to show decreasing spending in real dollars as retirees age (about 1% per year on average), they calculated that the 3.8% initial safe withdrawal rate would increase to 4.3% if expected increases in future spending were limited to 1.8% per year (2.8% annual inflation minus 1% each year). They also noted that if retirees adopted more dynamic approaches such as the Guyton/Klinger Guardrails approach, the starting withdrawal may be increased from 3.8% to 5.3% and still satisfy the 90% probability of success criteria over 30 years based on their assumptions.
Static vs. Dynamic Spending Strategies and Stochastic vs. Deterministic Assumptions
Static and Dynamic spending strategies are defined in our glossary, but essentially static strategies like the 4% Rule and static Monte Carlo models employed today by some financial advisors anticipate a “one and done” calculation of how much fixed real dollar annual spending can be supported each year by either a household investment portfolio (the 4% Rule) or by household assets (static Monte Carlo model) based on current information, assumptions about future experience and, generally, relatively high probabilities of success developed using stochastic assumptions. Under this type of strategy (which some refer to as a “faith-based” strategy), household spending is generally expected to increase each year with inflation irrespective of actual experience. While annual spending may be more stable under this type of strategy, a household runs the risk of either spending too much or too little. Static spending strategies can be made more or less conservative by varying the underlying assumptions about future experience, spending less than the safe amount or, as mentioned by the Morningstar authors, by assuming lower- than-inflation future spending increases.
Dynamic spending strategies, like the AFP and dynamic Monte Carlo models, anticipate periodic adjustment of withdrawal rates or spending based on actual investment and spending experience. Therefore, future spending may fluctuate, but it also can be smoothed to some degree. Initial spending under dynamic spending strategies can be made more conservative by using more conservative assumptions about the future or by spending less than the spending strategy amount each year (for example by building up a larger Rainy-Day Fund). As time goes by, however, dynamic strategies will adjust for actual experience. As noted above, the AFP model is a dynamic model and it uses deterministic assumptions, and therefore, the initial spending budget results from AFP can be considered as approximately equivalent to a 50% probability of success under a stochastic model utilizing the same assumptions about the future.
Comparison of Initial AFP withdrawal rates with Morningstar withdrawal rates
The AFP is designed to produce an Actuarial Balance Sheet and spending budget for a given set of assumptions, household assets and desired spending classified as recurring, non-recurring, essential and discretionary, but we can significantly simplify the asset and desired spending entries to compare initial withdrawal rates under the AFP with the Morningstar report results.
Let’s assume that we have a single male retiree age 65 named Bill. Bill has an annual Social Security benefit of $20,000 and accumulated savings of $1,000,000. His initial desired spending is about $67,000. Bill considers his essential spending to be $42,000 in real dollars with the rest (the spending that would bring his plan funded status to approximately 100%) discretionary. The screenshot below shows the AFP for Bill under default assumptions for this example, assuming that Bill invests about 56% of his accumulated savings in low-risk investments to approximately balance his non-risky investments with the present value of his essential expenses.
 |
| (click to enlarge) |
This screen shot shows that Bill’s initial expected withdrawal from his accumulated savings (spending of $66,500 minus Social Security of $20,000) is $46,500, or about 4.65% of his accumulated savings of $1,000,000. This percentage compares with Morningstar’s “safe” withdrawal rate of 3.8%. Thus, even though the default AFP assumptions about the future are somewhat more conservative than the Morningstar assumptions, the initial AFP withdrawal rate is higher than the Morningstar withdrawal rate of 3.8% because it is effectively based on a 50% probability of having to reduce spending in the future rather than the 10% probability used by Morningstar to develop their 2022 safe withdrawal rate.
If we had used Morningstar’s assumptions, and kept the methodology the same as described above (including the initial value of $42,000 in essential spending), the AFP would have developed an initial $78,000 spending budget and an initial withdrawal rate from accumulated savings of 5.8% [($78,000 - $20,000 Social Security)/$1,000,000]. This compares with the initial 5.3% withdrawal rate produced by the Guyton/Klinger guardrails approach, which is dynamic for the first 15 years of retirement, but static thereafter.
If we had used the same 4.5% investment return assumption for risky assets/investments (rather than the 7.5% default rate in the AFP and 9.5% assumed for Morningstar) and the remaining AFP default assumptions, we would have developed an initial spending budget in this example of $59,000 and an initial withdrawal rate of 3.9%; much closer to the Morningstar 3.8% safe withdrawal rate. We leave it up to our readers to see the effect of assuming inflation minus 1% future increases in their discretionary spending.
We conclude from these comparisons that while the Morningstar assumptions about future experience are more optimistic than the default assumptions used in the AFP, generally the static withdrawal rates in the Morningstar report are more conservative (lower) than initial withdrawal rates determined by the AFP since they are based on a 10% probability of having to reduce withdrawals in the future rather than approximately 50% for the AFP.
Do our results imply that the AFP is less conservative than the 4% Rule or static Monte Carlo models? In terms of initial spending, yes. Is this a problem? No, because the AFP approach is a dynamic strategy and is self-correcting over time. Because it is not static, there is no need to develop an overly conservative initial spending budget that is expected to be successful 90% of the time without adjustment. In our opinion, this makes dynamic approaches like the AFP more reasonable than static approaches. Our opinion is confirmed by Michael Kitces and his crew as discussed in the next section.
Kitces.com discusses static vs. dynamic strategies
In the Kitces.com post of December 21, 2022, Derek Tharp walks us through a discussion of static (one and done) approaches and dynamic (ongoing plan) approaches. Mr. Tharp concludes that the dynamic approach is generally superior to the static approach for planning and dynamic approaches are not terribly sensitive to probabilities of success (or failure) generated by Monte Carlo models. He says,
“…momentary probability of success is not a very intelligible concept when change is planned for from the outset, even to advisors who likely understand Monte Carlo simulation significantly better than most people.
Unlike the one-time plan where a lower probability-of-success level does meaningfully influence the risk of depleting a portfolio, lower probability-of-success levels have a trivial impact on the risk of depleting a portfolio if adjustments will be made going forward.
…this draws attention to an interesting disconnect between how advisors commonly think of probability-of-success thresholds. According to the common view, probability-of-success thresholds tell us something about the likelihood of depleting a portfolio at a given spending level. However, recall that this is only true for one-time projections that will not experience spending adjustments.”
Note that while Mr. Tharp suggests using a Monte Carlo model (with perhaps a 50% probability of success) updated on a periodic basis for planning purposes, the same (or better) planning results can be achieved with the AFP. And, unlike most Monte Carlo models, the AFP provides more flexibility when entering different types of anticipated expenses and different expected increases in those expenses and it permits entering more reasonable spending upon the first death withing a married household.
Changing assumptions in the AFP
As discussed above, you can make the AFP more conservative (lower initial spending) or more aggressive (higher initial spending) by using more or less conservative assumptions about the future. When doing this, remember to hit the “override” button in Column D of the spreadsheet and input the replacement assumption in Column F. Don’t change the assumption shown in Column I directly. If you do, you may need to download a fresh copy of the spreadsheet.
Conclusion
We are not big fans of the 4% Rule or other static spending approaches. However, the 4% rule is very popular, and it is worthwhile to periodically compare results of the 4% rule under a given set of assumptions about the future with results of other models, including the AFP. If you simply want to perform a one-and-done exercise at the beginning of your retirement and charge ahead each year with blinders on so that you ignore actual experience, a static approach may be for you. As an actuary, I believe the AFP is a much more robust planning tool than the 4% Rule or other static approaches.
Not getting enough discussion of retirement planning using the AFP? Be sure to check out our recent Advisor Perspectives articles.
Five Steps to Help Your Retired Clients - Articles - Advisor Perspectives
Three Strategies to Strengthen Household Balance Sheets - Articles - Advisor Perspectives
Wishing you and your households Happy Holidays and a Happy New Year
