By Ben Mathew, Guest WriterHow much to withdraw from our portfolio each year is a central choice that we face in retirement. It’s a consequential choice. Withdraw too much, and we risk running out of money. Withdraw too little, and we risk missing out on things that would have been meaningful to us.
But how do we decide how much is too much and how little is too little? We won’t know for sure until we have run the actual course of our retirement. But we don’t have the luxury of hindsight. We have to make our decisions with the limited information that we have, recognizing that the future can play out in many different ways. This is the basic problem of financial planning for retirement.
Withdrawal strategies give us a foothold to start thinking about this problem. They let us systematically calculate a reasonable spending target given our current circumstances and preferences. Whether it’s a 65-year-old starting retirement with a $5 million portfolio or an 85-year-old with $200,000 left, the strategy provides some clarity around how much to spend.
It’s unlikely that you will follow any systematic withdrawal strategy exactly—you may need to take out more in some months and less in others. But the calculation serves as a guidepost that will alert you when you’re consistently spending too much. And conversely, it will let you know when you can afford to spend more.
Safe Withdrawal Rates (SWR) and Amortization Based Withdrawal (ABW)
There are so many withdrawal strategies and so many graphs and tables and backtests flying back and forth that it’s easy to get confused and overwhelmed by it all. But it turns out that most of these strategies are based on just two underlying methodologies: Safe Withdrawal Rates (SWR) and Amortization Based Withdrawal (ABW). Understanding these two basic frameworks can help us evaluate and make sense of this complex space.
SWR was introduced by a financial planner named Bill Bengen in a 1994 paper, and it has since become popular in the financial planning industry. ABW is the withdrawal strategy coming out of the academic literature in economics. The model that looks at how an investor should save, invest, and withdraw from the portfolio is known as the lifecycle model. Key papers on this topic were published by economists Paul Samuelson and Robert Merton in 1969. Both would go on to win the Nobel Prize in Economics—Samuelson in 1970 for his fundamental contributions to economics and Merton in 1997 for his work on derivative pricing. While ABW has a sound basis in economics, it’s also a simple and natural approach that has been independently discovered and adopted by retirees looking for a practical solution for their own retirement.
Withdrawal strategies based on SWR include:
- The 4% Rule of Thumb
- Guyton-Klinger Guardrails
- Risk-Based Guardrails
- Kitces’ Ratcheting SWR
- Early Retirement Now Withdrawal Toolbox: Main
Withdrawal strategies based on ABW include:
- The Life-Cycle Model of Economics
- Total Portfolio Allocation and Withdrawal (TPAW)
- Required Minimum Distribution (RMD)
- Variable Percentage Withdrawal (VPW)
- Early Retirement Now Withdrawal Toolbox: CAPE-Based Rule
While SWR is currently widely used in the industry, it has several issues that lead to poor withdrawal recommendations. ABW is a sounder approach, and it has been the basis of withdrawals in the economics literature since the papers by Merton and Samuelson came out 55 years ago. Though ABW has been widely used in economics, efforts to translate the results for a wider audience have been limited, and the financial planning industry has been largely unaware of the method. But awareness of ABW’s advantages is increasing, and more retirees are starting to adopt it to solve the issues arising from SWR.
More information here:
The Silliness of the Safe Withdrawal Rate Movement
Safe Withdrawal Rates (SWR)
Let’s first look at how SWR works and see what the problem with it is:
Fixed SWR
The basic SWR strategy involves withdrawing a fixed amount each year (inflation-adjusted) so that the probability that you won’t run out of money before the end of retirement is sufficiently high—say 95%.
For example, using historical US stock and bond returns, you might conclude that for a portfolio with a 50/50 fixed asset allocation, if you take out 4% of the starting portfolio plus inflation adjustments, there is a 95% probability that it will last you through a 30-year retirement. If you start with a $1 million portfolio at age 65 and take out $40,000 each year plus inflation adjustments, there is a 95% chance that you won’t run out of money until age 95. So, you conclude that spending $40,000 per year with inflation adjustments is reasonable.
This is the type of analysis behind the popular 4% rule of thumb for withdrawals.
A major drawback of this approach is that it assumes fixed withdrawals. It’s modeling a retiree who withdraws the same amount year after year regardless of how the market performs. That is not how most people would behave—and for good reason. If the market does badly and their portfolio performs worse than expected, it would be prudent for the retiree to reduce their expenses and withdraw less. If the market does well and the portfolio performs better than expected, it would be reasonable for the retiree to take out more.
The model assumes that you will withdraw a fixed amount each year, but in reality, you will adjust your spending in response to how the market performs. This is a basic mismatch. Still, we don’t expect any model to match our behavior exactly. The important question is whether the model is useful. Does the calculation give us a good enough approximation of reasonable withdrawals? Can it serve as a useful guide?
Unfortunately, fixed SWR results don’t serve as a good enough guide to retirement spending. The fixed withdrawal mismatch turns out to be quite impactful and skews the withdrawal calculation significantly. To make retirement work with fixed withdrawals, the withdrawal amount has to be very conservative. Because the retiree is assumed to be irrationally inflexible through bad years, the strategy is forced to preemptively select an overly conservative spending amount.
The full extent of this conservative bias of the SWR methodology is not often appreciated because the 4% withdrawal rule itself is not that unreasonable for the start of a 30-year retirement. But that’s because the 4% rule was calculated using historical US stock and bond returns, which have done spectacularly well. If we use more moderate expected returns based on current stock and bond yields (which are lower than what prevailed historically), replace US return data with global return data, or account for longevity risk, we get significantly lower withdrawals.
For example, Anarkulova et. al. (2023) calculated an SWR of 2.26% using a dataset from 38 developed countries. People often look at these low withdrawal rates, conclude that they are implausibly low, and then blame the reduced expected return assumptions as being unreasonable. But the problem is not the reduced expected return assumptions. The problem is that SWR is inherently conservative. It produces low withdrawals from reasonable return assumptions because it assumes away flexibility.
To move beyond fixed withdrawals and create a more realistic withdrawal strategy that can provide better guidance, variable versions of SWR have been developed.
Variable SWR
Let’s look at two different approaches to making SWR variable:
Implicitly Variable SWR
An obvious way to turn fixed SWR into a variable strategy is to adjust the withdrawal each year by simply rerunning the fixed SWR calculation. Each year, the retiree enters the updated portfolio balance and the remaining horizon and calculates a new withdrawal amount that corresponds to their chosen target probability of success.
With a fully adjusting strategy like this, you will not run out of money. If the market does poorly, your withdrawal also falls and prevents your portfolio from running out. This means that your actual probability of success will be 100% regardless of the probability of success you use in the SWR calculation each year. So, the probability of success used each year to recalculate SWR has lost its original meaning as a measure of risk.
There is still risk—but the risk now is low spending due to poor market performance, not running out of money. To understand the risk, you need to look at the probability of different spending outcomes at different ages. Spending in the first year will be the same as fixed SWR. But each year after that, spending can go up or down. What does the range of possible spending outcomes look like? Here is a graph showing spending outcomes from a $1 million 35/65 portfolio that targets a 95% probability of success over 30 years, calculated using US historical data from 1871-2020:
The spreadsheet with these calculations is available here.
The dark line in the middle is the expected trajectory of spending. It shows what spending will be if the market does exactly as expected. Spending will be higher if the market does better than expected and lower if it does worse than expected. The shaded region around the expected spending trajectory shows the 5th to 95th percentile spending outcomes. This gives a sense of how much variability you can expect and how much you may need to adjust. The riskier the portfolio, the wider this spending range will be.
The graph shows that while spending in the early years starts out pretty conservative, it rises quite a bit throughout retirement. Expected spending increases from $39,000 in the first year to $180,429 in the last year. Even the 5th percentile outcome in the last year is well above the starting withdrawal. Compared to a fixed SWR strategy of taking $39,000 each year and leaving a lot of money unspent, the strategy allows for more spending overall.
As we had anticipated, explicitly introducing flexibility in spending addresses the overly low withdrawals of fixed SWR. Starting withdrawals are still low, but withdrawals ramp up in the second half of retirement. This also suggests that we could—if we wanted to—spend more in the early years. This can be done by reducing the probability of success used in the SWR calculation to something below 95%—say to 80% or 50%. That will increase withdrawals in early retirement and reduce it in late retirement. This clarifies the role of the probability of success in this process. It doesn’t refer to the actual probability of success (which is 100%), but it indirectly controls the range of spending outcomes across time.
But there is still a problem with this strategy. The problem becomes apparent when you look more closely at the expected spending trajectory. Notice how spending grows relatively slowly in the early years of retirement and then ramps up quickly toward the end. In the first five years, expected spending increases 12% from $39,000 to $43,751. But in the last five years, it increases a whopping 67% from $108,017 to $180,429. Why did that happen? It’s not clear why spending should grow slowly in the early years and faster in later years. It’s not something that we intended to get out of this strategy. It turns out to be an arbitrary artifact of the process that we used. It’s just a side effect.
But your expected spending trajectory should be at the center of your retirement planning. It should be the focal point, not a byproduct of the withdrawal strategy. There are several important tradeoffs that you need to make here, and it is something that should be under your direct control.
For example, some people find it useful to break down their retirement into three phases: the Go-Go years, the Slow-Go years, and the No-Go years. The Go-Go years is the first phase when you are relatively young and healthy and can participate in more activities and travel. You may have higher expenses during this phase to support an active lifestyle. The Slow-Go years is the middle phase where health starts to decline and activities become more restricted. But you are still independent and can live without support, so expenses may become lower. Finally, in the No-Go years, your health might decline to the point where you cannot live independently and need assistance. Medical bills and end-of-life care may raise your expenses again.
How should you allocate your spending capacity between these different phases with very different needs and objectives? These are the types of decisions that you will need to make as you construct your expected spending trajectory.
Here’s a broader set of considerations that you will need to think about as you decide on your expected spending trajectory:
- Consumption smoothing: This is the desire to spread our base spending somewhat evenly across time. Barring special expenses like a home renovation, a major trip, or end-of-life care, we usually want to evenly divide our spending. We don’t want to spend a ton of money one year and have very little left for another year. That implies a flat expected spending trajectory.
- Precautionary savings: Since future spending is uncertain, even if you would like to spend an equal amount in each period, caution dictates that you should spend somewhat less in early retirement—just in case the future turns out worse than expected. This is a natural response to uncertainty and is formally known as precautionary savings. This early underspending means that your expected spending trajectory should be upward sloping.
- Wanting it sooner: You may prefer to spend more in early retirement during the Go-Go years when you are in better health and can do more things. Or you might simply value spending sooner rather than later. This would make your expected spending trajectory downward sloping.
- Needing it later: You may have more medical expenses as you grow older and may need to allocate extra for end-of-life care during the No-Go years. This would make your expected spending trajectory upward sloping.
- Extra expenses: Things like loan payments, travel, and kids’ college tuition will require spending extra in some years. This makes the expected spending trajectory lumpy.
You will have to decide how important these different imperatives are to you and settle on an expected spending trajectory that balances these needs. There is no one shape that fits all. You may want a spending trajectory that is upward sloping, downward sloping, flat, U-shaped, or bumpy. The important thing is that you are in the driver’s seat and making an informed decision about the tradeoffs.
With variable SWR, you can indirectly control the trajectory in a limited way by adjusting the probability of success. If you target a high probability of success, you’ll spend less in early retirement and more in late retirement. If you target a low probability of success, you’ll spend more in early retirement and less in late retirement. So, you have some control over the overall slope of the expected spending trajectory. But whatever probability of success you choose, you are still limited by the set of trajectories that happen to emerge from the process. You will still have relatively slow growth in the early years and faster growth in later years.
Explicitly Variable SWR
Another way to create variable SWR strategies is to explicitly discard the fixed withdrawal assumption and specify that the withdrawals will be variable according to some rule. For example, you can specify a rule that says to reduce withdrawals by X% if the withdrawal rate rises above Y%. Then, you look at the probability of success of following this rule. The idea is to pick a variable withdrawal rule that produces a sufficiently high probability of success—with success still defined as not running out of money before the end of retirement—while also not letting spending get too low. Guyton-Klinger guardrails and Kitces’ ratcheting SWR fall into this category of explicitly variable SWR. Risk-based guardrails technically fall into this category, too, but share some aspects of implicitly variable SWR as well.
Because explicitly variable SWR strategies don’t adjust fully, the probability of success need not be 100%. And since withdrawals are variable, spending will depend on how the market does. To get the full picture, as we did for implicitly variable SWR, we have to look at the expected spending trajectory and the range around it. Different rules will produce different expected spending trajectories and ranges.
As noted earlier, the expected spending trajectory is a key consideration in evaluating whether a strategy makes sense. What is the expected spending trajectory of these strategies? It’s hard to say. That’s because they are not targeting a specific trajectory. Just like with implicitly variable SWR, it’s a byproduct of the process. You can run simulations to see what it looks like. But since the rule wasn’t designed to achieve a specific trajectory, it’s going to be ad hoc.
These withdrawal strategies have the same problem as before. The expected spending trajectory is a side effect of the process and not an intentional choice based on the retiree’s preferences. The shapes of the trajectories of explicit SWR are different from those of implicit SWR, but they are not any more meaningful.
More information here:
I’m Retiring in My Mid-40s; Here’s How I’ll Start Drawing Down My Accounts
Amortization Based Withdrawal (ABW)
Now, let’s look at ABW and see how it fixes this.
In ABW, we flip this around. Instead of the spending trajectory emerging as a side effect of the planning process, ABW creates the process to deliver the spending trajectory that you want.
To understand how this works, let’s start with a simple example: Suppose you have $1 million, plan for a 30-year retirement, and want to create a flat spending trajectory—i.e. spend the same amount each year. How much can you withdraw from your portfolio each year? If the portfolio is not growing, this calculation boils down to simple division: $1 million/30 years = $33,333 per year.
This is called amortizing the portfolio.
If the portfolio is growing, the calculation becomes a bit more complicated, but it can still be done with some high school math and a calculator: For example, if the portfolio grows 3% per year, we can calculate that withdrawals would be $49,533 per year. (The formula for this is P*r*f/(1-fn) where P is the portfolio balance, r is the growth rate of the portfolio, f is the discount factor 1/(1+r), and n is the number of years. You can also use the PMT function in a spreadsheet: PMT(r,n,-P,0,1).)
This is a flat withdrawal schedule. We can also get withdrawals to rise or fall by adding spending growth to the amortization. For example, if the portfolio grows 3% per year and you want spending to grow 1% per year, then withdrawals would start at $43,664 in the first year, rise 1% per year, and end at $58,270 after 30 years. (This can be calculated by replacing r in the previous formula with (1+r)/(1+g)-1 where g is the growth rate of spending.) You can amortize the portfolio to construct any withdrawal schedule you want like this. See the Bogleheads wiki on ABW for more on how to amortize a portfolio.
But so far, we have not considered risk. Risk means that the actual return may be higher or lower than expected. What happens if the actual returns are higher or lower than the expected return?
The amortization is not a once-and-done withdrawal calculation. Just like we used SWR to construct a variable withdrawal strategy, we use ABW to construct a variable withdrawal strategy. We simply re-run the amortization each year with the updated portfolio balance and remaining horizon and recalculate the withdrawal schedule. If the portfolio grows as expected, you will continue on the original withdrawal schedule. If it grows more than expected, the new amortization will shift your withdrawal schedule up, increasing your withdrawals above what you had originally scheduled. If the portfolio grows less than expected, the new amortization will shift your withdrawal schedule down, decreasing your withdrawals below what you had originally scheduled. Each year, your withdrawal schedule gets bumped up or down depending on whether the portfolio did better or worse than expected.
With this variable amortization-based withdrawal strategy, just like with variable SWR, we have an expected spending trajectory and a range around it. The problem with SWR was that the expected spending trajectory that it generated was a side effect of a process that focused on other things. What is the expected spending trajectory of ABW? It is simply the original amortization schedule. That’s because if the return comes in higher than expected, withdrawals will be higher than originally scheduled. And if the return comes in lower than expected, withdrawals will be lower than originally scheduled. The expected withdrawal is what you had originally scheduled in the amortization.
This gives you full and direct control over the shape of your expected spending trajectory. To construct the trajectory that you want, all you have to do is amortize your portfolio accordingly. Whether you want a spending trajectory that is upward sloping, downward sloping, flat, U-shaped, or bumpy, you can do it all directly through the amortization. There is no reason to let it happen indirectly like in SWR.
For example, suppose we want an expected spending trajectory that grows 2% per year. We can use ABW to create precisely that. Here are the spending outcomes from amortizing a $1 million, 35/65 portfolio with 2% spending growth, using the same historical data as the SWR graph earlier:
The spreadsheet with these calculations is available here.
Withdrawal starts at $49,182—which is 26% more than the $39,000 given by the implicitly variable SWR strategy that was graphed earlier. Expected withdrawals then grow at a constant 2% per year, as scheduled, ending at $64,397 in year 30. Actual withdrawals come in higher or lower than the expected withdrawal depending on whether the market did better or worse than average.
Interestingly, even though withdrawals start 26% higher than the variable SWR schedule, the lowest 5th percentile outcome is $28,416 in year 19, which is 12% higher than the lowest 5th percentile outcome with variable SWR ($25,444 in year 14). That’s because variable SWR featured a relatively low and flat trajectory in the early years, which allowed the 5th percentile to drop lower as the range increased over time. The ABW schedule starts higher and grows at a constant rate, so it can keep the 5th percentile from dropping as low in the middle years. A relatively balanced spending profile like this makes an expected spending trajectory with a constant growth rate a good starting point for thinking about how to structure your retirement spending.
While ABW is a general method that allows you to construct any expected spending trajectory that you want, specific withdrawal strategies that use ABW may not offer all possible trajectories. The basic lifecycle models of economics have expected spending trajectories with a constant growth rate that balances the desire for consumption smoothing, precautionary savings, and a preference to consume sooner rather than later.
TPAW is a planning tool that I developed to help implement the lifecycle model. Since the basic lifecycle models feature trajectories with a constant growth rate, TPAW allows for a constant growth rate of spending. In addition, to allow more control over the timing of spending, it has options to vary the spending growth rate over time, add extra expenses in some periods, etc. The RMD withdrawal schedule is obtained by amortizing the portfolio assuming a real growth rate of 0% for the portfolio and a flat spending trajectory. VPW assumes a real growth rate of 1.9% for bonds and 5% for stocks and also uses a flat spending trajectory.
Given the variety of ABW implementations available, you have to make sure that the withdrawal strategy you use offers the trajectory that you want. If you can’t find an ABW-based strategy that suits your needs, you can always do the amortization yourself using a spreadsheet. Spreadsheet templates are available from the Bogleheads wiki on ABW.
More information here:
Fear of the Decumulation Stage in Retirement
A Framework for Thinking About Retirement Income
The Bottom Line
Most retirees will want to pursue a variable rather than a fixed withdrawal strategy. A withdrawal strategy should be evaluated in terms of how well it can provide guidance around variable withdrawals—how much can the retiree spend to begin with and how should they adjust their spending throughout their retirement?
What ultimately matters to the retiree is the distribution of spending outcomes—the expected spending trajectory and the range of outcomes around it. ABW provides a straightforward process that puts you in direct control of your expected spending trajectory, allowing you to shape it in any way you want. This lets you balance your need for consumption smoothing, precautionary savings, extra spending, etc. while arriving at a spending trajectory that relates to your preferences in a meaningful way. With SWR, you give up control over this central aspect of retirement planning.
How have you thought about spending in retirement? What strategies have you considered? Would SWR or ABW (or something else) be right for you?
Ben, this was awesome, simply awesome! There is such an emphasis on SWR, but I guess it’s hard to use ABW for planning purposes, correct? I mean, there is no 4% rule or similar rule to define FIRE using ABW, right? You can’t use ABW to make a retirement savings goal at least based on my reading, even if you had a trajectory goal of spending like the U shaped spending trajectory, right? But ABW seems a great way to maximize spending the way a retiree would want while at the same time decreasing the variability of spending.
My retirement plan with my wife was to die with zero. do you find that ABW would be the best way to do that? Or do you think Guyton-Klinger is the best way? I would need to know the shape of my spending trajectory, correct? Should I just assume the U shaped spending based on the research that most likely I will be using that type of spending?
Guyton-Klinger is just another in a long line of methods that start with SWR and then try and algorithmically improve upon it. There are as many of those out there, some of which are on the ERN site. At their heart, though, there is still SWR and for which there is still no guarantee that your money will last as long as you want it to.
With ABW, what I did before retirement was to do a “pretend” retirement to see what my withdrawals would be if today was the day I was going to retire. Because ABW results in variable withdrawals, you need to make sure you have sufficient margin for that variability. But once I saw I had the margin I was comfortable with, I gave notice and retired. At any time, if you’re short, you can easily compute how much more you need in your portfolio to meet your goal. I would recommend doing this periodically.
With ABW, you can target any terminal value you want for your portfolio. It’s pretty common to use $0 as the terminal value (for the die with zero case), but even then, people add margin by making the terminal age something fairly large.
Best
DC
Yes. I’ve generally used the average of life expectancy and 120.
I don’t recall who suggested that, but it works for me.
We can construct simple rules of thumb like the 4% rule using ABW. But once we start conditioning the withdrawal rate on time horizon, asset allocation, expected returns etc., it becomes easier to just use the PMT function on a spreadsheet and calculate it directly. Here’s an example table of ABW based withdrawal rates for those who don’t want to use a spreadsheet:
https://www.bogleheads.org/forum/viewtopic.php?p=7697928#p7697925
But a spreadsheet will be much more customizable and not much harder to use. The basic calculator in the ABW wiki should be pretty easy to use: https://www.bogleheads.org/wiki/Amortization_based_withdrawal
And an online calculator like TPAW Planner will also be easy to use:
https://tpawplanner.com/
Given these options, I don’t think it’s necessary to rely on rules of thumb.
Re: Guyton Klinger, as DC said above, it’s basically SWR being tweaked to make it variable. The better approach is ABW, which is simpler and clearer than these variable guard-rail based SWR methods.
The spending trajectory is a personal choice. I think assuming extra expenses for end of life care makes sense. But whether to spend more in early retirement would come down to personal preferences and circumstances.
Ben, thank you for this post. I’ve seen your TPAW on Bogleheads for years now, but never dived into studying it. I’m sure you have a good answer for my concern of how to go about choosing an amortization schedule when I really don’t know how long I’ll live. Your example mentions a 30-year retirement, but what if the person lives 40 years after only planning for 30?
To deal with uncertain lifespans, we can simply update the time horizon in the ABW calculation based on remaining life expectancy. Each year, the goalpost would move to the new life expectancy plus some buffer. The RMD approach does this. It amortizes the portfolio assuming zero real expected returns and a horizon based on remaining life expectancy.
TPAW does not model uncertain life expectancy yet, but we will implement it eventually.
This seems very complicated. What would be wrong with the following withdrawal strategy:
1. On years when the market is down, only take out 3-4% of your portfolio
2. On years when the market is up, take out the total return minus maybe 3% or so to add to your portfolio. So for example if market is up 10% one year, I take out 7% and then keep 3% in the portfolio.
3. For years it is just a 3-5% return, take out the full amount and you can add an additional 1-3% if you need (almost like it was a down year).
Seems a little simpler and allows you to take out way more in good years to enjoy, but still living off 3-4% in bad years.
I think what you are proposing is more complicated that amortizing the portfolio, and also wouldn’t work as well. Amortization is just splitting up your portfolio over the remaining years. There is a little bit of math in splitting it up, but conceptually it is simple and easy and behaves in predictable ways.
One problem in the method you suggested is that taking x% of the portfolio would too aggressive when the horizon is small (early in retirement) and too conservative when the horizon is short (in late retirement). At a minimum we should be talking about x% for a particular horizon.
As for varying the withdrawals based on up/down years, taking out the total return when the market is up will lead to large increases in spending in those years. The gains of the up years should be spread over the remaining years. And the losses of the down years should also be spread over all remaining years. That is effectively what amortization would do, which will cause it to generate smoother spending outcomes. Spending will be variable, but not more variable than it needs to be because we’d be always spreading out the impact of unexpected gains and losses over the remaining horizon.
I used the SWR for the first 6 years of my retirement. It was safe and less complicated than almost anything, and gave me training wheels while I was adjusting to the scary process of beginning to draw down my retirement savings. By the end of my highest risk period, I was more comfortable with the process and saw that the RMD schedule seemed to be a good match for my spending, and easy to use.
Now, I run my calculations three ways (designed in anticipation that one or more of my calculators would fail or no longer be supported, which has already happened) but the RMD seems to be the strongest guidepost.
I’m very happy to see the math backing up a decision I accidentally backed into.
Thanks for sharing your experience. Glad to hear you switched from SWR to RMDs and find it to be a better guide. RMD does a lot of things right. The main issue is that it’s very conservative, reducing spending in early retirement. If that’s to your liking, you’re all set. If you want to be less conservative and spend more early, you can always re-do the RMD amortization using a lower spending growth rate, shifting spending more into the early years.
More about the RMD method in this post:
https://www.bogleheads.org/forum/viewtopic.php?p=7389497#p7389497
The spreadsheet linked at the end of that post can be used to adjust the RMD amortization:
– If you enter withdrawal growth (g) = rate of return (r), you get the RMD amortization.
– If you enter withdrawal growth (g) > rate of return (r), you will get a more conservative schedule.
– If you enter withdrawal growth (g) < rate of return (r), you will get a less conservative schedule.
For some reason, the last part of my reply is not posting properly. Trying words instead of symbols:
– If you enter withdrawal growth (g) equal to the rate of return (r), you get the RMD amortization schedule.
– If you enter withdrawal growth (g) higher than the rate of return (r), you will get a more conservative schedule.
– If you enter withdrawal growth (g) lower than the rate of return (r), you will get a less conservative schedule.
Excellent article! I appreciated the very clear explanations and links for additional details/reading. I was, however, surprised by this statement: “Though ABW has been widely used in economics, efforts to translate the results for a wider audience have been limited, and the financial planning industry has been largely unaware of the method.” I’ve never worked with a financial planner, and I guess before I do, I should find out exactly what they can and cannot offer me.
I think it’s important to spend a little more time on part of this article that a lot of people just skipped right over at the beginning:
It’s a consequential choice. Withdraw too much, and we risk running out of money. Withdraw too little, and we risk missing out on things that would have been meaningful to us.
It’s not a consequential choice to many successful investors. They have enough money that they neither risk running out of money nor missing out on meaningful things. This stuff only matters when you barely have enough or want out just as soon as you can get out or absolutely want to spend every dime you can. I don’t think that’s nearly as many people as many authors of articles like this assume.
But for that segment of people who want to retire as soon as they can/spend as much as they can, this is a pretty important topic.
Good point Jim. I should spend a little more time on that statement but just superficially, even though I didn’t read Die With Zero yet, I fear the “Withdraw too little, and we risk missing out on things that would have been meaningful to us.” Being screwed by Northwestern Mutual as created a scarcity mindset where, even before I got suckered into whole life, i was tight, but post losing $50,000 to whole life, I am even more TIGHT! I am scared to be the richest man in the grave, and knowing my psychological weakness I want to spend every last freaking dollar/have my last check bounce to enjoy first class/buy a ferarri and gun it at 200mph/sip Mai Tai’s in the Maldives before it sinks/watch the Superbowl live (preferably with the Giants playing, although not sure any amount of retirement saving will accomplish that goal).
I guess they say money can’t buy happiness, but it’s the relationships that matter, so maybe I will also pay for my son to accompany me with flying first class/buy a ferarri and gun it at 200mph/sip Mai Tai’s in the Maldives before it sinks/watch the Superbowl live. I like Ben’s approach here because it really does help me accomplish this goal.
Agree that people who save much more than what they’ll possibly need to fund their retirement spending don’t need to model their withdrawals carefully. Their portfolios are so large and their spending needs so little that no math is necessary. They will leave a lot of money behind when they die and are perfectly happy with that.
There are definitely some people in that situation. Not only do they not need to think hard about withdrawals, they also don’t need to think hard about asset allocation. They’ll be fine with 100% stocks, 100% bonds or anything in between because the cushion is so large. Tax planning won’t matter except to their heirs. Traditional vs Roth vs taxable is not going to move the needle on their spending. There will be enough money for their needs as long as they don’t invest rashly.
For people who are not in that situation—for those who are faced with a real choice between spending more today vs spending more tomorrow, between taking less risk vs getting more reward—for those people, it’s useful to think carefully about how much to save, how to invest those savings, and how to withdraw from their portfolio in retirement.
TPAW and ABW calculators I’ve got from B’heads don’t take into account taxes or various types of retirement savings – taxable, tax free, pretax.
It seems to me these are important variables in terms of calculating post portfolio withdrawals, as well as expected levels of growth.
Also, presumably one would rerun this every year to validate the spending level for the year coming up.
If that’s the case, it’s not clear to me how using this is substantively different to a risk-based retirement spending guard rail approach.
– Agree that modeling taxes is important. It’s on the to-do list for TPAW. I expect it to be implemented sometime in 2025.
– Yes, you would re-run ABW with the updated portfolio balance, horizon, etc. The difference between a variable withdrawal strategy generated by re-running ABW vs re-running SWR is described in the post starting with the Variable SWR section. ABW works better because you can understand and control the expected spending trajectory well. With SWR, the expected spending trajectory becomes opaque and poorly behaved.
My method is simple: I take my five year average return – not a generic return, the five year average inflation rate, the number of years I have left until I reach 100 years old and then compute it using a calculator on mycalculaltors.com / retirement % withdrawal. I redo this every year. https://www.mycalculators.com/ca/retcalc2m.html
It may go up, it may go down. I could drop dead tomorrow or live to be 101 – I will “worry” about that in a few years.
I’m a UK resident, 4 years into retirement and have been something of a disciple of ERN and SWR for some years.
I’ve always been a little troubled by how the traditional SWR approach leads to a very high probability (perhaps 95%) of leaving a significant amount unspent at the end of life. This problem was brought into sharp focus last October when the UK Government changed the Inheritance Tax threshold to include the value my pension pot, which would previously have gone to my children free of tax.
I came across TPAW last year and it is no exaggeration to say that it has been a life changer! Thank you Ben.
I have set the tool so that it gives me more to spend now (on travel), reducing at 70 then again at 75, and (importantly) leaving me with a reasonably large balance at 80. Assuming I make it to that age, I can at then buy an annuity to cover my remaining days.
The result is that I have significantly more to spend now, and I won’t leave anything to the avaricious tax man.
Glad you’re finding TPAW helpful in constructing a spending plan that reflects your personal circumstances and preferences!
SWR has to be overly conservative because we’re assuming away flexibility. A variable amortization based withdrawal strategy that aligns spending with the portfolio, adjusting along the way as needed, better captures the nature of the withdrawal problem for most retirees.
Ben, I have been refining my use of your tool this afternoon. I notice that in setting it to leave me a “legacy” at 80, the recommended equity allocation is increased significantly towards the end, because the tool assumes a higher risk tolerance for the legacy. I have therefore amended my approach so that instead of a “legacy”, I set a single month’s “extra spending” of £500k on my 80th birthday. This adjusts the equity allocation down to finish at 0% on my 80th birthday, which works much better.
One thing that perturbs me slightly is that my portfolio’s current 65% equity allocation is described by the tool as a “very aggressive” risk level; even 31% equity is described as “aggressive”. I know this is largely a subjective judgement, but it doesn’t seem to follow conventional thinking. Perhaps you can explain.
Many thanks
You can change the risk tolerance for legacy in Risk -> Advanced -> Increase Risk Tolerance for Legacy. If you move that slider to 0, then legacy will have the same asset allocation as general spending.
The risk tolerance labels of “conservative”, “moderate” and “aggressive” refers only to the risk portfolio and does not reflect essential expenses which get funded by 100% bonds. In other words, if you have categorized a lot of expenses as essential and funded them by 100% bonds, thereby creating a floor for those expenses, then even an “aggressive” allocation on the remaining risk portfolio would not be aggressive in terms of spending. The risk to spending is what is shown in the spending graph. Adjust the risk tolerance slider to arrive at the spending graph you like the most. If that aligns with your preferences, you don’t need to worry about the “conservative” vs “moderate” vs “aggressive” labeling on the slider.
Thanks Ben. I’ve followed your suggestion and reverted to the “legacy” approach dialling down the risk tolerance to zero and hey presto, it works much better!
Flexibility is highly valued in life, but particularly in retirement. The more flexible you can be, the more you can spend.
Ben;
To be clear, the TWAP planner tool you have created does NOT take into cost of taxes? So the amount of spending would included taxes? So, my broad assumption is that my taxes will be 24% initially and rise to 35% in later years do to RMD and larger income.
Congrats on a fantastic retirement if you’ll be paying 35% either as a marginal rate or an effective rate.
Yes, taxes have not been modeled yet in TPAW. We will incorporate it eventually. For now, you can use the following approach as an approximation:
– If most of your savings is in traditional accounts, enter pre-tax income and consider the monthly spending and savings shown in the graph to be pre-tax.
– If most of your savings is in Roth accounts, enter post-tax income and consider the monthly spending and savings shown in the graph to be post-tax.
– If most of your savings is in taxable accounts, enter post-tax income, reduce the expected return to account for the tax drag, and consider the monthly spending and savings shown in the graph to be post-tax.