By reviewing existing research on variable spending, we can identify and describe key representative variable spending strategies from the countless possibilities, and classify them into a general taxonomy. I identify three general categories for variable spending strategies:
- decision rule methods,
- actuarial methods, and
- dynamic programming methods.
Key examples from the first two categories are shown in Exhibit 1. The third category, based on complex dynamic programming computational methods, moves beyond our scope of coverage and is not widely used in personal finance.
That being said, I will cover dynamic programming methods within the context of how they compare to decision rules and actuarial methods after first discussing these methods.
Exhibit 1: Variable Spending Strategies in Retirement
|Decision Rule Methods
|Bengen’s Constant Inflation-Adjusted Spending
|Endowment Formula: Weighted Average of Methods (1) and (2)
|Endowment Formula: Fixed-Percentage of Three-Year Moving Average Portfolio Balance
|Bengen’s Dollar Floor-and-Ceiling Withdrawals
|Vanguard’s Percentage Floor-and-Ceiling Withdrawals
|Kitces’ Ratcheting Rule
|Guyton and Klinger’s Decision Rules
|Zolt’s Glidepath Spending Rule
|Modified Required Minimum Distribution (RMD) Spending Rules
|Apply PMT Formula (with different returns, longevity, and spending smoothing)
|Monte-Carlo PMT Formulas:
|Frank, Mitchell, and Blanchett’s Age-Based 3D Model
Blanchett, Maciej, and Chen’s Mortality-Updating Constant Probability of Failure
Blanchett’s “Simple Formula”
Though numerous exceptions exist, we can try to generalize a few important distinctions for these methods. Among these, decision rule methods frequently share elements of the probability-based school of thought, while advocates of actuarial methods (and dynamic programming methods) often identify more with the safety-first school.
Decision rule methods will demonstrate more willingness to start spending at a higher level than justified by the bond yield curve, with an expectation that future portfolio growth from stocks can be counted on to justify a higher spending rate now. Meanwhile, with actuarial methods, spending may start at a lower level, and spending will only increase in the event that upside potential has been realized.
There is a greater recognition of the notion that stock investments are still risky even after long holding periods, so efforts to “amortize the upside” through higher spending may backfire.
Decision rules will generally try to keep spending at a steadier level and only make spending adjustments when deemed essential, while actuarial methods may call for more frequent spending adjustments.
Actuarial advocates suggest that those seeking smoother spending should at least use a less volatile portfolio. For actuarial methods, spending volatility is more directly linked to investment volatility.
It should be used to reduce spending volatility, rather than any other sort of smoothing technique. However, some actuarial methods may seek to smooth spending as well. Any effort to keep spending constant from a volatile portfolio creates risk for even greater subsequent spending declines.
Beyond these differing views of market risk, decision rule methods will generally adopt a conservative planning horizon beyond life expectancy (such as thirty or forty years), while actuarial methods will make decisions based on a dynamically adjusting time horizon linked to the remaining life expectancy as retirement progresses.
Finally, decision rule methods will generally be more comfortable in formulating their spending parameters using historical market data, whereas actuarial methods will be more willing to incorporate updated market return expectations as the spending plan is updated regularly throughout retirement.
With this overview in mind, I plan to go into greater detail on each strategy in upcoming posts.
To find out more about creating a retirement plan that works for you, read our eBook 6 Steps to Creating Your Retirement Plan.