Kudos to Michael Kitces and Derek Tharp for attempting to fix some of the deficiencies in spending models typically used today by financial advisors, as previously discussed in our post of July 23, 2020. In their post of March 3, 2021, they highlight some of the problems with Strategic Withdrawal Plans (SWPs) and Monte Carlo models typically used today by financial advisors, and they propose incorporating the guardrail concept for determining annual withdrawals in SWPs advocated in the “Guyton-Klinger Rule” into Monte Carlo “Probability of Success” models to enable financial advisors to better advise their clients. And while we believe the resulting “Probability-of-Success-Driven Guardrails” (or Kitces/Tharp) approach is definitely an improvement over current practice, we remain unconvinced that it is superior to the Recommended Financial Planning Process advocated in this website.
Kitces Post
If you or your financial advisor use either a SWP or Monte Carlo modeling to develop your spending budget, we encourage you to read “Using Probability-Of-Success-Driven Guardrails To Manage SafeRetirement Spending.” In this article, the authors note several of the deficiencies associated with SWPs (with or without guardrails) and typical Monte Carlo approaches that we have noted in many of our prior posts. They then build a case, and provide an example, for incorporating spending guardrails in typical Monte Carlo models.
In their conclusion, Messrs. Kitces and Tharp state:
“a probability-of-success-driven guardrails approach captures the communication advantages of traditional guardrails approaches without overlooking the client-specific cash flow, longevity, and other nuances that are captured in Monte Carlo analyses but that are overlooked by traditional guardrails approaches.
By bringing these two frameworks together, advisors can deliver higher-quality plans for their clients that can also be more effectively communicated to clients.”
Where the Kitces/Tharp Model Still Falls Short
The primary shortfall of the Kitces/Tharp model, in our opinion, is its inability to distinguish between recurring and non-recurring expenses and between essential and discretionary expenses. Instead, like most Monte Carlo models employed today by financial advisors, it produces one fixed dollar spending budget for the current year. If client spending goals include spending on non-recurring expenses such as mortgages, travel costs, family assistance, new cars, kitchen remodeling, etc., these specific client spending goals are ignored. Treatment of non-recurring expenses as recurring results in inefficient use of the client’s limited assets. And while it may be reasonable to assume that total spending will decline over time, assuming declining real-dollar spending on essential expenses will generally not be prudent.
We have included below the functionality comparison shown in our post of July 23, 2020. We note that while adding guardrails to the typical Monte Carlo model would provide useful information to clients to facilitate decisions on when to increase or decrease future spending, it does not address most of the other functionality items we noted in that post.
Functionality Comparison
Item | Kitces/Tharp Model | Actuarial Financial Planning Process |
Permits entry of different assumed rates of future expected increases for different types of future expenses? | No | Yes |
Permits expenses to be reduced by X% when first in a couple is expected to die? | No | Yes |
Permits entry of different lifetime planning periods for different members of a couple? | No | Yes |
Automatically adjusts plan spending for actual experience? | Yes | Yes |
Reflects all assets (including home equity) and spending liabilities in model? | No | Yes |
Distinguishes between non-recurring and recurring expenses? | No | Yes |
Distinguishes between essential and discretionary expenses? | No | Yes |
Quantifies value of non-financial assets? | No | Yes |
Quantifies size of Floor Portfolio necessary to fund essential expenses? | No | Yes |
Conclusion
As discussed many times in our website, pension actuaries don’t generally use Monte Carlo approaches (with or without guardrails) to develop contribution requirements for their clients. Using deterministic models and annual valuation processes generally enables a pension actuary to more easily value more complicated plan provisions and communicate results effectively from year to year. Most pension plan sponsor clients were not interested in possible contribution ranges or probabilities of success associated with a given contribution level. We believe similar deterministic models and annual valuation processes can work very well with individual retirement plans.
Messrs. Kitces and Tharp do a pretty good job in their article describing some of the communications problems associated with using Monte Carlo models and some of the weaknesses of SWP approaches. These concerns about Monte Carlo modeling were also raised by James Sandidge and discussed in our post of January 29, 2021. Instead of communicating potentially confusing probabilities of success, Mr. Sandidge recommended that financial advisors communicate results of stress tests. We included a stress-testing example in that post to illustrate how one can go about using our ABC workbooks to illustrate how a client’s plan may perform under adverse investment conditions.
We applaud Messrs. Kitces and Tharp for their efforts to help financial advisors better advise their clients. As we have suggested to these gentlemen more than a few times in the past, they may also wish to look at the Recommended (Actuarial) Financial Planning Process advocated in this website as an approach to help financial advisors deliver even higher quality plans that better meet client-specific spending goals.
