[EDITOR'S NOTE: Here at The White Coat Investor, we know our readers love having real-life examples of portfolios and how people accumulate their money and then eventually spend it. That's why we want to hear from those who have already retired and who are living their lives in a post-work world, so those of us who are still working can be inspired and learn how to get where you are right now. Please fill out this form and inspire us with your wisdom. Don't worry, we'll keep your identity a secret. We plan to take your answers and create even more content for those who want to learn about how to spend in retirement. Help us help others!]
Have you ever looked at a Monte Carlo analysis of sustainable retirement spending and thought to yourself, “How could there be so much variability in potential outcomes?” That might especially be true given that the same asset allocation (or mix between stocks and bonds) drives the results.
A $20.8 Million Dispersion in Outcomes
Here is an example of a 64-year-old couple with a $4 million portfolio (60% stock/40% bond) who plans to retire at age 65 and spend an inflation-adjusted $172,000 after-taxes per year from Social Security and portfolio income and withdrawals:
At a 99% probability of success rate (which is far too high, by the way), the variability of outcomes is striking. The worst-case actual historical observation (having started this retirement in November 1936), indicated by the red line, shows this couple would have passed at 98 years old with $1.22 million (real, inflation-adjusted dollars). The best-case actual historical observation (having started this retirement in July 1982), indicated by the green line, shows this couple would have passed with $22.03 million. This is a $20.81 million difference in possible outcomes based on actual historically experienced sequences of returns.
The standard of living that could be afforded in retirement under these different outcomes is wildly different. While we cannot know in advance the path of returns we will be blessed with during our retirement years, your choice of withdrawal strategy in retirement can have an enormous impact on your dispersion in outcomes and the quality of life you can lead while retired.
More information here:
The 4% Rule and Safe Withdrawal Rates
4 Methods of Reducing Sequence of Returns Risk
Total Return with Rebalancing (the Conventional Way)
The example presented illustrates a “total return with rebalancing” withdrawal strategy. It is, by far, the most prevalent strategy used by financial advisors and DIYers alike. This strategy chooses an asset allocation (typically a mix of stocks and bonds) based on a retiree’s goals and risk profile. Systematic withdrawals (e.g., monthly) from the portfolio, often based on maintaining a “safe withdrawal rate,” are funded by a combination of income, capital gains, and principal. The portfolio is rebalanced back to the target asset allocation on a pre-determined basis (quarterly or annually) by selling investments that have grown in value and reinvesting the proceeds in the investments that haven’t.
Total return with rebalancing is an easy withdrawal strategy to implement and delivers a predictable retirement “paycheck” because it is often based on a static withdrawal rate. It also captures the benefits of rebalancing, a forced mechanism to sell high and buy low consistently—one of the hardest things to do as an investor.
Despite its benefits, this approach has several significant shortcomings:
- It overly relies on asset allocation as the lever to control the sequence of returns risk (see our previous WCI guest post entitled “The Folly of Relying on Asset Allocation to Manage Sequence of Returns Risk”).
- It requires conservative asset allocations paired with static “safe” withdrawal rates, which tend to lead to underspending in retirement. Whether the couple in our example ended life with $1.22 million or $22.03 million, they never enjoyed spending or gifting more than $3.59 million from their portfolio while living.
- The steady, inflation-adjusted retirement “paycheck” approach does not mesh well with retiree spending research, showing that inflation-adjusted spending tends to decline over the course of retirement (J.P. Morgan Asset Management, David Blanchett, and more). In this case, steady spending leads to less available spending early in retirement when you're healthy and can enjoy it most and it provides more than needed later in retirement.
- It relies on “probability of success” as its measure. Is an 80%-90% probability of success suitable? What does it mean to fail? Is any probability of failing retirement even really acceptable? The point is that total return with rebalancing attempts to find the maximum safe withdrawal rate corresponding to the probability of success you’ve chosen and locks you into that level of spending, assuming you can’t (or won’t) change over a 30+ year retirement regardless of what actually has happened or is happening. The higher the probability of success you choose, the lower your withdrawal rate and the more you possibly leave on the table upon passing. This is why a 99% probability of success would be considered way too high. A 99% probability of success would have to set your sustainable spending so low that it could pass even the worst-case scenario (even though it is not probable).
- It fails to incorporate reality as it unfolds. If you found yourself on a great path of returns during the first 10 years of retirement and had twice what you started with, wouldn’t you want to increase spending and/or gifting? We have a lot of clients who have a secondary goal of gifting to their children once they know they are not at risk of becoming a financial burden to them. If we experienced a second Great Depression, would you not curtail your spending—at least some—out of fear?
The total return with rebalancing withdrawal approach is a great way to provide a stable cash flow in retirement and to maximize the amount of inheritance for your heirs, but it’s not the best way to maximize the joy you could derive from your wealth while living—whether it's through travel, shared experiences, or gifts you get to see enjoyed.
Bucket Strategy (or Time Segmentation)
The bucket strategy allocates money based on asset-liability matching, which matches future investment sales with planned expenses. Generally, there are three “buckets,” which can be separate investment accounts. As with all strategies, there’s leeway in how they’re designed.
Here is an example of how a retiree with a $3 million investment portfolio could allocate their account if they planned on withdrawing $150,000 annually.
The near-term bucket would have the first three years of retirement spending invested in stable investments, like a money market fund and Treasury bills. The intermediate-term bucket would have the next three years of spending in investments that provide income and growth, like bonds. And the long-term bucket would be designated for your spending needs seven years out and beyond, invested aggressively for growth in stocks and real estate.
A significant benefit of bucketing is that it’s easy to understand and it plays nicely with our tendency to mental account. This helps retirees feel comfortable spending, which can be challenging during early retirement and heightened market volatility. It’s also customizable based on preferences (e.g., the near-term bucket can be extended to five years of spending for the risk-averse), and having a near-term bucket helps mitigate the sequence of returns risk.
But does bucketing help retirees avoid making unnecessary sacrifices? We don’t think so. Bucketing is really just total return with rebalancing in disguise. If we add the buckets in our example, we get a 70% stock/30% bond asset allocation. So, bucketing strategies suffer the same shortcomings as a total return with rebalancing withdrawal strategy.
Additionally, bucketing adds complexity to the rebalancing decision, as rules need to be created to “refill” the buckets. If buckets are not refilled, your investment portfolio becomes increasingly aggressive as you spend down your near- and intermediate-term buckets. And since rebalancing calls for selling growth assets to buy stable assets, some of the benefits of traditional rebalancing, where assets are sold based on how they have appreciated in value relative to other portfolio assets, can be lost.
Dynamic Withdrawal Strategies (or Flexible Spending Rules)
Dynamic withdrawal strategies create spending rules designed to change retirement income distributions based on actual market performance. Dynamic withdrawal strategies solve for the shortcomings presented by the total return with rebalancing and bucketing strategy approaches. Dynamic withdrawal strategies allow a retiree to start with an initial withdrawal rate exceeding the “safe withdrawal rate” allowed in the others because mid-course corrections can and will be made during bad and good markets as needed or allowed. Dynamic withdrawal strategies also provide comfort that a retiree won’t run out of money in retirement because changes will be required long before that can happen.
The tradeoff is that dynamic withdrawal strategies require a retiree to accept some level of variability in their retirement income. How much variability a retiree is willing (and able) to accept will be determined by an individual’s situation and the extent to which fixed expenses are covered by non-portfolio income sources (i.e., Social Security, pension, etc.). The good news is that most dynamic withdrawal strategies are customizable, and they can fit income variability and floor constraints.
The most popular form of dynamic withdrawal strategy uses rules to set upper and lower “guardrails” around some measure, typically withdrawal rate or portfolio value. The upper guardrail serves as the “prosperity rule,” when an increase in your withdrawal is warranted because the portfolio value exceeds the upper guardrail. The lower guardrail serves as the “preservation rule,” when a decrease in your withdrawal is required because the portfolio value fell below the lower guardrail. These guardrails keep a retiree on track and serve as signals when an increase in retirement income is warranted or when a decrease in retirement income may be necessary. Dynamic withdrawal strategies help retirees consume their portfolios more efficiently because they factor in both portfolio performance and spending (not just spending).
More information here:
Fear of the Decumulation Stage in Retirement
A Framework for Thinking About Retirement Income
What the Research Shows About Retirement Drawdown
Morningstar provides some insightful findings in its research paper, “The State of Retirement Income: 2023,” as researchers modeled the safe retirement drawdown rate for 2023 (i.e., total return with rebalancing strategy) and then compared it to five other retirement withdrawal strategies. A summary of their key findings based on a 40% stock/60% bond portfolio over 30 years at a 90% probability of success rate follows:
The “Base case” method represents the conventional total return with a rebalancing and inflation adjustment approach. The “Guardrails” method is based on the popular Guyton-Klinger approach (more on this later). Morningstar’s research found that the guardrails dynamic withdrawal strategy provides the highest starting safe withdrawal rate at 5.2%. The guardrails approach provides 30% more retirement income than the total return with rebalancing approach, which over 30 years means a much higher quality of life and level of enjoyment. The cost for this higher starting withdrawal rate is a lower median Year 30 ending value (at $800,000 vs. $1.5 million) and greater cash flow volatility. Still, this tradeoff is well worth it for the vast majority of retirees we work with who would rather spend more while living and who can enjoy it or gift it while they can still see the benefit rather than having wealth transfer at death (even if a little more wealth could be transferred).
Dynamic withdrawal strategies are one of the most powerful tools retirement planners have in their tool chest. Vanguard, in its “Advisor’s Alpha,” and Morningstar, in its “Alpha, Beta, and Now . . . Gamma” research papers, both point to the material value of dynamic withdrawal strategies. Morningstar even goes so far as to say dynamic withdrawal strategies can create more value than all of the tax-management strategies combined (fortunately, these things aren’t mutually exclusive because we believe in both strongly and apply both in practice).
Expanding on our earlier example of a 64-year-old couple with a $4 million portfolio (60% stock/40% bond) who plans to retire at age 65 and spend an inflation-adjusted $172,000 after-taxes per year from Social Security and portfolio income and withdrawals. Following is a summary of the effects of raising their spending target to $220,000 after-taxes per year and the further effect of then overlaying on a dynamic withdrawal strategy:
The probability of success goes from dropping to 69% to improving to 99% (recall, dynamic withdrawal rules will require changes far before risking failure). Most importantly, cumulative withdrawals made during retirement are the highest under any scenario. The variability of outcomes narrows substantially, because you are spending/gifting more during retirement when you can.
More information here:
I’m Retiring in My Mid-40s; Here’s How I’ll Start Drawing Down My Accounts
Guardrails Strategies in Practice
If a guardrails strategy is so great, why doesn’t everyone use it? Short answer: they are challenging to implement in practice. Most guardrail rules are based on withdrawal rates (including the most popular guardrail strategy, Guyton-Klinger). Withdrawal rate-driven guardrail strategies are problematic because they rely on steady portfolio withdrawal rate patterns that seldom occur in reality. Your portfolio withdrawal rate (especially in earlier retirement) can jump around based on your sources and needs for retirement funding. For instance, perhaps you are two years away from claiming Social Security, so you need to take more temporarily from your portfolio until benefits kick in (i.e., take a higher withdrawal rate). Perhaps you are executing a Roth IRA conversion strategy requiring greater funding to meet temporarily higher tax liabilities. Of course, less would be needed later as tax liabilities drop. Withdrawal rate-driven guardrails, like Guyton-Klinger, don’t play nicely with these external realities.
The good news is that risk-based guardrails are an alternative approach that solves the withdrawal rate-driven guardrail shortcomings. The bad news is that it is even harder to implement without access to specialized retirement planning software. Shifting from withdrawal rates to risk provides a more holistic approach to spending adjustments that allow us to apply guardrails to the type of unique spending profiles we typically see in reality.
While a complete examination of risk-based guardrails exceeds the scope of this article, an example of a risk-based guardrail strategy could look as follows:
- Initial Withdrawal Rate: Begin by spending an amount that has an 80% probability of success.
- Upper Guardrail: If the probability of success rises to 100%, increase spending to a level with 20 points higher risk (80% probability of success).
- Lower Guardrail: If the probability of success falls to 25%, decrease spending to a level with 20 points lower risk (45% probability of success).
If your goal in retirement is to make the most of your assets while living so you can enjoy them, a risk-based guardrail dynamic withdrawal strategy will be the right choice for you.
What are your thoughts on a retirement withdrawal strategy? What would be your preferred option? Would you rather spend more retirement money now or leave more of it behind for your heirs and charities?
So, the bottom line for implementing the risk based withdrawal is to each year re-adjust your spending (up or down) so that you’ll have an 80% projected success rate going forward. The question is, what formula or model do we use for calculating that 80% success rate? Or is that something only paid advisors can tell us. I have Boldin (New. Retirement) and it spits out success probabilities under several scenarios. Could one theoretically use one of those scenarios and keep using that scenario each year to recalculate the success rate?
This is one issue with complicated withdrawal strategies. Now imagine continuing to implement it (or your spouse doing so) as mental faculties diminish.
Fair enough, although it strikes me that a guardrails-type approach to decumulation (maybe especially a risk-based rather than Guyton-Klinger one?) will probably be particularly helpful in avoiding underspending during the “go-go” years. Straightforward enough to downshift to something simpler/more spouse-friendly as one heads into the “slow-go” and especially “no-go” phases.
Anything that gets people to actually spend money they can afford to spend is probably a good thing. Funny that these tools are what it takes for some to do that eh? We’re funny people, humans.
I think this is what the author has in mind, and it seems to me that Boldin’s—or in my case Empower’s—modeling is sophisticated enough, assuming you are synced up with precision, as I am.
Frankly I was blown away by this idea which I had not encountered before. I think this modeling software is already really good, and is likely to just get better. I had been adjusting my plans to the “retirement success” needle for years already in the accumulation phase. Why didn’t it occur to me before to use the same guardrail in the withdrawal phase? Seems much easier than the floor&ceiling method I am using, and way easier than the CAPE method I was toying with.
Would love to hear from someone who is tightly integrated with Empower who thinks this is a bad idea.
I’ve been a big fan of risk-based guardrails ever since first coming across them in a great blog post from Michael Kitces a couple years ago. He’s written about it a couple of times, most recently in March 2024: https://www.kitces.com/blog/guyton-klinger-guardrails-retirement-income-rules-risk-based/
One of the most important points Kitces makes in this article is that advisors using risk-based guardrails should change the way they talk about it with their clients. It’s not measuring probability of success, per se. It’s measuring the probability that a change in spending will be required. So if MC shows a 70% “success” rate, it really means there’s a 70% chance that spending at the initial level or better with inflation will persist throughout the retirement, and there’s only a 30% chance that a reduction in spending will be needed. That’s a major change in mindset from 70% chance of success.
I still have the same question as Sanjay about which tool to use to get the Monte Carlo probability. I’ve been using https://ficalc.app, a free webapp, and https://www.firecalc.com/ along with a proprietary MC tool my investment my paid advisor offers. Also, just downloaded FRP, https://www.flexibleretirementplanner.com/, and will start experimenting with that.
The goal is to find a tool that I can use consistently over time. Ease of use, likelihood of long-term availability, and withdrawal strategy tunability are my criteria for selection.
My wife and I are fairly close to the example couple used in this WCI article in terms of age, portfolio size and retirement stage. We are nowhere close to the spending level we could afford, so for now the MC question is somewhat moot. Working on changing that and upping our spending, which is harder than it sounds.
Your last paragraph is so common among WCIers that I actually think it will apply to a solid majority of them eventually. Which makes any sort of complicated withdrawal strategy seem silly. It just doesn’t matter what you do when you’re withdrawing 2.5%.
By comparison I am on a knife’s edge! $2.5M portfolio with a $80-100K burn rate! As Seneca said, if you set a high value on liberty, you must set a low value on everything else.
I guess by comparison. But I wouldn’t lie awake at night even if I had a $2.5M portfolio and was spending $120K a year. Historically, 5% works fine for 30 years 67-82% of the time. 5% of $2.5M is $125K. Spending usually drops after the go-go years anyway and most people retiring at typical retirement ages don’t live 30 years. So for this all to come crashing down you have to have all of the following go “wrong”:
# 1 Have a higher than 4%ish withdrawal rate
# 2 Run into a bad sequence of returns
# 3 Live a long time
# 4 Be incapable of adjusting to the new reality if all of the above occurs.
What are the odds? Not that high. Live a little.*
*Note that I climb mountains for fun. Investing for retirement is one of the least risky things I do.
Thanks Jim. Been coming here for years, and still finding new wisdom.
I like this perspective.
#1 4% is very manageable,
#2 sequence of returns is manageable with adequate cash/short term TIPS/Treasuries for 7-10 years,
# 4 yes, we can adjust,
#3 long life – who knows what the future will hold. I’m close enough to the base scenario described that I think I should up the spending for at least the next decade.
Thanks TomS. Those two links to ficalc and firecalc look to be very useful. I’ve bookmarked them and hope that with those two and Boldin, I’ll be getting 3 independent ways to assess safe withdrawals.
I just got a license to Flexible Retirement Planner (https://www.flexibleretirementplanner.com/wp/), a Monte Carlo-based retirement planning tool. All I had to do was make a small donation via PayPal and I got my license key ($10 – one time).
Based on my first run, it is showing a 94% success probability based on our current portfolio at $4.6m and initial WR of $160,000. It has a neat goal-seeking function you can use to tell it what success rate you want to hit and it will return the corresponding WR.
Since we’re using a risk-based guardrails approach I am targeting 70% success rate with a $100k floor. With those inputs it came back with a WR of $220k – very much inline with WCI’s example in the above article.
So far I think I like this MC simulation the best of any of the ones I’ve tried. It is clean and has just the right amount of tunability without being overly complex.
***** UPDATE – MAY 2025 *****
Around three months ago I obtained an individual license to IncomeLab, a planning platform designed for CFPs to use with their clients. It is built around the Risk-Based Guardrails methodology for determining and adjusting annual retirement “income” amount (e.g., the amount you can spend each year from your portfolio).
So far, I like it far better than any of the other tools I’ve seen. No more wondering whether I should aim for a 90% or 95% success rate. No more wondering whether I’ll have enough money left over to donate a new wing to my university’s engineering school, or will be broke.
RBG is the way to go, and IncomeLab how I’m doing it.
I’ve been interested in Income Lab as a tool to implement Risk Based Guardrails. I am not a financial advisor but am interested in Income Lab for myself. My current financial advisors use Right Capital, which at some point in the future will likely have this capability. Fortunately, my advisor is willing to plan with the Risk Based Guardrails, calculated manually. I set him the Kitces article and he’s forwarded it up to his management team.
How were you able to obtain an individual license and at what cost?
Thanks!
They don’t advertise it. I emailed them at their contact address with a different question related to this, and they came back to me and said that although they don’t offer it to individuals generally, they can provide an individual license for $20 a month, which is a continuation of their trial rate, with the understanding that it will be self support only, no training or 1:1 sessions. Email them and ask about it.
Thanks! Sorry, just seeing this now.
Enjoyed learning more about different withdrawel methods. Thanks Anthony for writing this article. Do you or Jim know of a good video explaining the different withdrawal methods? I would like to watch one that is credible and recommended by someone like you or Jim. I still have about 12 years before I start withdrawing money from my retirement accounts (Traditional, Roth, Brokerage), but I want to learn the best withdrawal method now so I can plan accordingly. Thanks!
First of all, there is no “best”, at least no consensus on exactly what is best. And there is likely a different best for different people. Many retirees, at least among those who read blogs like this one, retire with so much money it really doesn’t matter how they withdraw it. Like there is no best asset allocation, there is no best withdrawal method. They all have their pluses and minuses. You need to find a reasonable one that works for you.
Thanks Jim for your reply. I should have worded my previous comment better. I meant a video that explains the different withdrawel methods so I can make an educated decision on which method will be best for my situation. I have noticed some videos only explain a couple of withdrawel methods, thus providing only limited options, which does not allow me to make an informed decision. I agree many retirees who read this blog and others like it won’t run out of money no matter what withdrawel method they choose, however many of those same retirees are likely highly educated and likely desire to pick a withdrawel method that results in the least amount in taxes. In this case, which method one chooses does matter. I will very likely be one of those that it won’t matter which method I choose, but my desire and passion for education still leads me to find the best withdrawal method for my situation.
Oh, you’re asking a different question than this post is answering and there probably IS a right answer to your particular question. I’m not the world’s best video maker (I’m far better at blogging than any other medium IMHO) but I think you’re looking for this:
https://www.whitecoatinvestor.com/how-to-spend-in-retirement/
Most of these discussions of withdrawal methods like this one are really talking about how much to take out each year, not where to take it from. That link above discusses that and I think that’s what you’re most interested in today.
Yes, that is what I am most interested in. I read the article (how to spend in retirement) and it was helpful. Thanks!
I find the article and comments a bit misleading.
First, the bullet points in “Total Return with Rebalancing” section look wrong to me:
1. I can’t parse the part about “overly relying”. Asset allocation, rebalancing, etc. – these are all tools with various knobs. The specific knob values are the result of simulations and finding the global maximum of a function you optimize. Any strategy will rely on something. How do we define “overly” here? Bucket strategy is asset allocation strategy, after all.
2. Is 80-85% stock conservative allocation? That’s what usually various simulations show to be the best allocation for 30+ year horizon.
3. RE retiree spending trend: it’s just one of the possible inputs when considering ANY strategy. Nobody prohibits accounting for it (or maybe other way to say it – let’s do apples-to-apples comparison!). Even basic https://firecalc.com/ can account for it (in the form of Bernicke’s Reality Retirement Plan). More so, you can and should account for other things like SS, pension, etc.
4. Same as the next point – there is no rule that you should not adjust based on new information. ALL these strategies, at the end of the day is just guesses based on limited data. It’s easy to see it by doing a very simple thought experiment – imagine that it’s 1965 right now. The Great Depression looks like the worst sequence you need to account for. History is just about to present you with a surprise in the form of 1965-1980 sequence – sequence with worse sequence of return!
5. If you found yourself on a great path of returns during the first 10 years – just do the recalc, and if your numbers tell you can use 2x more money, use it. Same in reverse. It’s nonsense that any strategy works better for the future – it’s all based on past performance.
For everyone thinking about implementing any type of strategy like that, I’d recommend exploring relevant articles by ERN, for example:
– https://earlyretirementnow.com/2017/02/15/the-ultimate-guide-to-safe-withdrawal-rates-part-10-guyton-klinger/
– https://earlyretirementnow.com/2023/06/16/flexibility-swr-series-part-58/
At the end of the day there is no free lunch – you trade one set of problems for another.
Consider using real historical data (sequences), not Monte-Carlo simulations.
Some related reading – see https://earlyretirementnow.com/2018/01/24/random-walk/, “Implication 3”
And for the record – I love ERN articles for their data driven / scientific approach. But I also run my own simulations, using real historic data (same way ERN does it), accounting for things like SS, retirement spending smile, etc.
We have assets outside of our retirement plan (real estate, private investments). I would appreciate an article on how incorporate those into a spending plan.
Thanks!
I’m not sure why you think the location of the assets matters as far as a spending plan. It really doesn’t. Have you read this article? I’d start here:
https://www.whitecoatinvestor.com/how-to-spend-in-retirement/
That’ll tell you how to spend/what to spend first etc. We have lots of other ones, like this one, that talk about how much to spend.
Have not. Will read it