How do Safe Withdrawal Rates compare to Dynamic Retirement Income?

1 C Retirement Distribution ProcessThis post will briefly discuss how Safe Withdrawal Rate (SWR) research compares to my joint research with collaborators which contributes to the Dynamic retirement income approach and school of thought which adjusts retirement income if need be (the other school of thought being Safety First which SWR fall into).

The safe withdrawal rate (SWR) approach doesn’t explain how, or when, to transition to different time frames. The dynamic approach is actually built upon these transitions annually by definition and design and, as you will see below, it can also explain where and how the SWR approach intersects within the dynamic view.

This shouldn’t be interpreted as saying one is better than the other – simply to say that there is a method of comparison. The objective of this blog post is to show how both schools of thought fit together mathematically to form a spectrum within which to make a more informed decision about prudent retirement income.

Basically the SWR approach looks at historically poor market returns sequences and answers the question: What is the safest income withdrawal rate a person could start with during the worst of times and not run out of money before they die? Listening to a presentation in San Francisco by Michael Kitces, there were basically 3 time periods with extraordinarily bad economic and market circumstances. Should a person retire at any of those times, the safest rate to withdraw money for retirement income going forward from those time periods was 4% (hence the 4% rule name). All other historical time periods could have higher withdrawal rates – which means the 4% rule is ultra conservative.

This blog basically seeks to answer the questions: “To what portfolio balance would the portfolio value need to decline to before a retiree should consider cutting their spending?” “How does this reduced value compare to the 4% rule?” “How does the reduced value work as the retiree continues to age?”

In order to compare SWR approach to the Dynamic Updating* approach (my term that expresses reviewing this dashboard and updating it with current information each year during the annual review) we need to look at such an example. Let’s take a look at the below dashboard which illustrates a hypothetical couple both age 60 wanting to begin withdrawing from their total $1,000,000 (note all dollar values may be scaled; e.g., $500,000 would be half of the $1,000,000 dollar values in this example). The left side of the dashboard shows where their holdings are kept within their Long Term Portfolio (equity allocation 25%) and their retirement income is coming from the Distribution Reservoir portfolio (all cash and short term bonds) going into their check book. Expected longevity for this couple is 29 year (upper right box).


The center and right side of the dashboard shows the current Long Term Portfolio value (green bar) and also shows what kind of decisions should be made should the markets misbehave and portfolio values decline to certain pre-calculated decision points. I blogged earlier on these, what I call, Emergency Procedures – how to make non-emotional decisions during emotional times.  On this point, also see Kitce’s Do you REALLY have a PLAN for dealing with a decline in the markets? I can’t state enough how important it is for retirement income success, to have rationally determined decision points and what decisions to make at those decision points, i.e., Decision Rules.

So, under a dynamically updating process, this couple could withdraw approximately $46,400 this year, or $3,867/month which is 4.64% (shown in dashboard that corresponds to a 10% Possibility of Adjustment (POA)) (which means a 90% Possibility of Excess).

Now, for the sake of comparison, let’s skip down to the 25% POA line (red bar) we see both a portfolio value of $760,025 with an adjusted monthly income of $3,477 (which means that the income should be adjusted to this value, IF the Long Term Portfolio value ever reached $760k) and the Possibility of Adjustment may dynamically revert from 25% back to 10% POA (retiree’s choice to adjust or not), the same POA now with spending held at the higher $3,867/mo based on starting Long Term Portfolio value of $860,800 (the balance of the total $1,000,000 is held separately in a different allocation in the Distribution Reservoir portfolio). How likely might this be? How necessary is it to constrain spending based on a lower portfolio value based, even though our real portfolio value is higher?

One could get a sense of this by looking at how much the Long Term Portfolio value would need to decline to reach $760k from $1 million – about 11.7% (actually, dashboard shows changes to the Long Term Portfolio value is used to make these decisions … so that value to make this decision is $860,800). The Standard Deviation of this hypothetical portfolio for this couple is 6.45% – which means that two-thirds of the time, market fluctuations would be that or less (11.7 divided by 6.45 = 1.8 Standard Deviations … or almost 2) (The 68-95-99.7% rule). Meaning it is possible, but it would take conditions not often seen – less often than two-thirds of the time (1 Standard Deviation).

Why do I discuss Standard Deviation here? To give a sense of what SWR means to you taking money from your portfolio. What if I just started with taking $3,477 per month from the total $1,000,000? This would mean I would not likely need to reduce my spending when portfolio values go down – values would need to go below $760k by definition here. $3,477 times 12 is $41,724 per year … compared to $1,000,000 (41,724/1,000,000) = 4.17% … not far from the 4% rule ($40,000/yr or $3.334/mo). What is the Possibility of Adjustment taking out 4.17% relative to the original $1 million? A little over just 2%, in other words, 98% possibility of excess accumulations. In other words, money is probably being left on the table for heirs by not being comfortable spending it as the retiree.

So what this means, by applying the SWR (4% rule), is that you have established a constrained, lower monthly income you are unlikely to need to reduce. However, the price for doing that is the difference between that and a dynamically adjusting income figure – or $390.00 per month ($3,867 – $3,477). The $3,867 also may not need to be reduced, but does have a possibility that you may have to (remember the 10% possibility of adjustment?) … it depends on what markets might do – beyond anyone’s control – but exposure to those market moves is within one’s control.

Please do not interpret the 25% Possibility of Adjustment usage here to mean that is the actual exposure to risk. The actual exposure to risk is 2% POA. Recall that I backed into what value the portfolio would need to drop to for that 25% POA possibility to exist based on a higher spending amount ($3,867). Then I used that lower portfolio value to determine the lower spending amount ($3,477). By using the 25% POA concept, we may actually calculate the lower spending amount that would not need to be adjusted should portfolio values decline. Indeed, with the actual portfolio value of $ 1million, and the lower spending amount ($3,477), the actual POA is just over 2% – thus, very close to the 4% rule rate which presumes a near 0% POA. The advantage?

This calculation process automatically adjusts as retirees grow older – where the spending rate diverges more and more from the 4% rule (because the remaining time period gets shorter – thus allowing for higher spending rates for those shorter periods.). For example, when this couple reach age 75, they have approximately 16 years expected remaining longevity at that age (age 91), with a corresponding withdrawal rate of about 7.04% (as compared to the 4.64% above)**.

In summary, the 25% POA corresponds to the higher spending ($3,867) and the lower Long Term Portfolio value ($760,025). The 2% POA corresponds to the lower spending ($3,477) based on the higher Long Term Portfolio value  ($860,800). In both cases, the Distribution Reservoir portfolio contains the other approximate $139,200 (Refer to the picture for this blog). Both of the spending amounts are derived from the 10% POA in Monte Carlo simulations (10% of the simulations run out of money and do not reach the end year) with the higher spending amount from the higher portfolio value, and the lower spending amount from the lower portfolio value. They are used to get an idea of what a prudent high spending rate may be subject to adjustment on one end, and a safe spending rate not subject to adjustment on the other end of the decision spectrum.

Moral of the story: you may need to adjust income down a little using a Dynamically Updating method, but you also don’t run the risk of accumulating “a bajillion dollars” (Michael’s technical term) by setting income based on being too safe (strict application of the 4% rule). The Dynamic Updating method can mathematically compare itself to the SWR as I’ve demonstrated above. The choice is yours along the spectrum between the two.

Safety First may view the need to adjust spending as a plan failure. Dynamic Updating views not spending what may be prudently spent when it could be spent as a plan failure. Again, there’s a spectrum between the two views. Oftentimes, market fluctuations would expand and narrow the spectrum differences.

Note also that the dashboard would be dynamically updated each year so the withdrawal rate percentages would slowly increase because withdrawal periods are slowly decreasing (life expectancy slowly gets shorter). Finally, the 4% rule works for long periods of time, i.e., young retirement ages – it doesn’t apply to older ages with shorter distribution periods remaining as the example above showed.

Note, through reference to the research data available through the links within each of the referenced papers, that the dynamic process ends up with an ending balance once those time periods actually arrive (there’s still a portfolio balance – although market sequence uncertainty means nobody knows what that balance might be – greater or less than needed). This uncertainty is what worries some. However, in the quest to gain certainty, income potential is lost by being a bit too safe. Balancing either approach through mixing them is prudent. And … recognize that life has, and always will have, uncertainty.

Some may say this is complicated. No more so, in my opinion, that understanding breakeven for refinancing a mortgage, the differences between the two Roth 5-year rules, or nuances of Social Security claiming strategies for example.

Disclosures: The above discussion should only be used with full assessment of your data that reflects your specific information and situation. Results will vary based on risk and return characteristics of the portfolio in question, ages and period life table used.

*Dynamic Updating goes beyond an annual review for the client. It includes updating expected longevity age based on the now new current age (and using period life tables as they get updated), updating portfolio values and allocation, and updating the risk and return characteristics of the data series for their allocation (adding the past year’s data to prior data), to finally run a new series of Monte Carlo simulations using all of the updated factors. This process slowly converts, year by year, past uncertainty into updated data and slowly the uncertain time period grows shorter with age as time remaining shortens. Of course, the future is always uncertain – the objective is to prudently manage uncertainty through a pre-determined decision making and updating process.

**By calculating what the 25% POA portfolio values and spending amount at age 75 (and each age prior to, and after) the Dynamic Updating method basically translates, for sake of near comparison, what the 4% rule may be at the given age. In other words, a spending rate that has a low POA for the specific expected time remaining.

, , , , ,

11 Responses to How do Safe Withdrawal Rates compare to Dynamic Retirement Income?

  1. Larry Frank, Sr. July 22, 2014 at 1:30 pm #

    In reference to the Distribution Reservoir in the graphic above … I Would also point out that cash doesn’t necessarily mean just cash (see the referenced research paper quote below)… it could use short term bond funds with very low volatility mixed with cash that is needed for the very short term. You get a bucket strategy anytime you divide the total retirement pie into separate pieces regardless of how those pieces are called. For example a bond ladder would be one of the buckets, although not a cash bucket.

    “The primary conclusion of this study is that a two-bucket strategy incorporating
    cash reserves along with an IP has an impact on the probability that clients
    will be able to meet their retirement goals. The results suggest that the
    long-held financial planning belief that a portfolio should have a cash reserve to
    mitigate the risk of having to sell investments at depressed price levels and
    mitigate excessive taxes and transaction costs should continue to influence
    retirement planning strategies”. … The Benefits of a Cash Reserve Strategy in Retirement Distribution Planning by Shaun Pfeiffer, Ph.D.; John Salter, Ph.D ., CFP®, AIFA®; and Harold Evensky, CFP®, AIF®

  2. Larry Frank, Sr. August 25, 2014 at 1:42 pm #

    Why not use a simple spreadsheet for retirement calculations? Here’s a great blog post why not

    And this tutorial online also explains why

  3. Larry Frank, Sr. September 4, 2014 at 8:14 am #

    My blog post on how all the research papers we’ve done fit together and what each one looked at along the research spectrum

  4. Larry Frank, Sr. September 5, 2014 at 1:24 pm #

    And another good article on how spreadsheets and simple average returns lead to different results than may be experienced

  5. Larry Frank, Sr. October 20, 2014 at 6:49 am #

    From and email exchange with Michael Kitces this morning:

    From the slide share numbers – lower left corner in slide –most of the time monthly income may be as high as $3867/mo for the upcoming year (it would adjust in subsequent years based on facts known then). However, should markets decline then the WR% would rise with a rise in the signal from the related POF (one can’t tell when the WR becomes a problem – the POF is a clear signal related directly with WR). If one foreshadows spending retrenchment and pre-calculates what the floor may need to be, in this example $3477/mo (far right in box) … then this floor value becomes the Necessary expenses target, and the amount between the floor and the upper value is the discretionary amount ($390 – or 11% of necessary).

    You can then see that 89% of expenses are really 100% of the necessary category and these won’t need to be messed with (the blog demonstrates how this value relates to the Safe Withdrawal Rate – SWR mathematically) typically because these relate very closely with the SWR value. Thus, 11% of the expenses should be, by definition, discretionary, and may be spent or not spent depending on what has happened during that particular year. So as a practitioner, and a deeper insight into measuring and monitoring boundaries, it then becomes important to understand what the client’s situation is so that one who under-saved is not inadvertently put into the position of having to cut spending when they’re already pushing the limit constrained by the lower spending value. Those who have saved sufficiently, or even over-saved, typically don’t face issues (gets to Meir Statman’s “Precarious Middle” demographic ).

    Humbly, I think this dynamic method actually is what you’re seeking for how to determine adjustments and transitions off the SWR floor. Quoting a segment in an email exchange with Wade earlier this year: “So I think you’re on an interesting track thought process wise. However, a refinement to experiment instrument design may be needed (physics again where instruments designed to measure the experiment influence results unless carefully designed not to). The main point so far is that WR% alone is a POOR method for comparison of data. Why? From our [David and my] JFP June 2010 paper: To represent all the variables incorporated into a withdrawal percentage value, a more complete withdrawal rate expression would be WR% N, I, i, @, μ, t, Δt where, for the portfolio, N = number of distribution years of the portfolio simulation, I = inflation, i = rate of return, @= standard deviation, μ = probability of failure, t = time set at a fixed target end date, for example age 95, and Δt = change in time (years) to represent aging of the person. There are two “time” functions to represent the number of years for the portfolio, and the number of years of the person. “

    We [Mitchell, David and I] found that POF (μ component) is a decent signal for better decision making.

  6. Larry Frank, Sr. November 11, 2014 at 2:49 pm #

    My comment on Wade Pfau’s blog explains another method to determine a floor (25% POF) dollar amount and ceiling (10% POF) dollar amount of spending given your present portfolio value and allocation characteristics.

  7. Larry Frank, Sr. November 22, 2014 at 8:04 am #

    Specific statistics about standard deviation may be found here … the 68, 95 and 99.7% rule

  8. Larry Frank, Sr. April 19, 2015 at 8:50 am #

    Robert Powell summarizes a paper by Wade Pfau that compares difference strategies. My dynamic updating method is mentioned by Wade, but the modeling is more indepth that the simple models he evaluated (so Wade did not include the model in his comparison). Here’s a link to Robert’s description which has a link to Wade’s paper.

Leave a Reply