Study on SWR rates for Early Retirement
Study on SWR rates for Early Retirement
I just came across this study from Early Retirement Now :
https://earlyretirementnow.com/2016/12/07/theultimateguidetosafewithdrawalratespart1intro/
In it, he essentially replicates the Trinity Study but looking at investment horizons out to 60 years. The results and discussion span several posts and I must say, the discussion is superb!
Take away points:
1) When looking at a 60 year investment horizon, a SWR of 3.253.5% is more appropriate.
2) Wheras the Trinity study showed that an equity allocation between 50100% was reasonable for the 30year period, this study shows that for 60 years, equity allocations in the 70100% range are necessary.
3) Starting valuations (as measured by CAPE) matter.
https://earlyretirementnow.com/2016/12/07/theultimateguidetosafewithdrawalratespart1intro/
In it, he essentially replicates the Trinity Study but looking at investment horizons out to 60 years. The results and discussion span several posts and I must say, the discussion is superb!
Take away points:
1) When looking at a 60 year investment horizon, a SWR of 3.253.5% is more appropriate.
2) Wheras the Trinity study showed that an equity allocation between 50100% was reasonable for the 30year period, this study shows that for 60 years, equity allocations in the 70100% range are necessary.
3) Starting valuations (as measured by CAPE) matter.
The Espresso portfolio: 

16% LCV, 16% SCV, 16% EM, 8% Int'l Value, 8% Int'l Sm, 8% US REIT, 8% Int'l REIT, 20% Interterm US Treas 

"A journey of a thousand miles begins with a single step."
 TheTimeLord
 Posts: 4667
 Joined: Fri Jul 26, 2013 2:05 pm
Re: Study on SWR rates for Early Retirement
Just showing the table from the article.
Run, You Clever Boy!
Re: Study on SWR rates for Early Retirement
^^^ TheTimeLord is linking directly to the image which is fair use for copyright. See: Linking to Copyrighted Materials (Inline linking). I'll help with the attribution.
Source: The Ultimate Guide to Safe Withdrawal Rates – Part 1: Introduction, from earlyretirementnow.com.
Source: The Ultimate Guide to Safe Withdrawal Rates – Part 1: Introduction, from earlyretirementnow.com.
Re: Study on SWR rates for Early Retirement
TheTimeLord wrote:Just showing the table from the article.
And those probabilities are actually still too optimistic. When taking into account a) today's high CAPE ratio and b) a higher than zero final portfolio value target the 4% becomes less and less viable:
https://earlyretirementnow.com/2016/12/21/theultimateguidetosafewithdrawalratespart3equityvaluation/
It's time to rethink the safe withdrawal arithmetic when retiring early!
Re: Study on SWR rates for Early Retirement
DDM4inv79 wrote:When taking into account a) today's high CAPE ratio and b) a higher than zero final portfolio value target the 4% becomes less and less viable
Here are two articles which call into question the continued usefulness of the Shiller CAPE10 as a measure of the richness of stock market valuations:
http://internationalfinancemagazine.com ... Bias.html
http://www.etf.com/sections/indexinves ... edcontext
Re: Study on SWR rates for Early Retirement
DDM4inv79 wrote:TheTimeLord wrote:Just showing the table from the article.
And those probabilities are actually still too optimistic. When taking into account a) today's high CAPE ratio and b) a higher than zero final portfolio value target the 4% becomes less and less viable:
https://earlyretirementnow.com/2016/12/21/theultimateguidetosafewithdrawalratespart3equityvaluation/
It's time to rethink the safe withdrawal arithmetic when retiring early!
No, they don't.
The probabilities are what they are. The question you need to ask is if the metrics of the economy today share commonality with the failures vs. the successes? Understand why the failures are failures. Simply put, what is different or unique about *today* that matches the kind of failures observed in the Trinity/new study?
Re: Study on SWR rates for Early Retirement
czeckers wrote:I just came across this study from Early Retirement Now :
https://earlyretirementnow.com/2016/12/07/theultimateguidetosafewithdrawalratespart1intro/
In it, he essentially replicates the Trinity Study but looking at investment horizons out to 60 years. The results and discussion span several posts and I must say, the discussion is superb!
Take away points:
1) When looking at a 60 year investment horizon, a SWR of 3.253.5% is more appropriate.
2) Wheras the Trinity study showed that an equity allocation between 50100% was reasonable for the 30year period, this study shows that for 60 years, equity allocations in the 70100% range are necessary.
3) Starting valuations (as measured by CAPE) matter.
But, thank you for this study. I care mainly about points 1 and 2  folks usually talk about a longer retirement requiring a 2% "SWR", which I thought was absurdly low. The data seemed to point to 30 years being equivalent to infinity. This points out that 30 years in not infinity, but it's pretty close, and the numbers generated pass the "sniff test" for myself, both in sustainable rates as well as AA required to support those rates.
The key takeaway (confirming my bias!) is that growth is needed even in the retirement phase to support retirement in later years / as a hedge for living longer.

 Posts: 24
 Joined: Fri May 20, 2016 11:00 am
Re: Study on SWR rates for Early Retirement
At what age(s) is it considered early retirement? I'm 61. Is anything less than 65 "early?" Just curious.
Re: Study on SWR rates for Early Retirement
They use data from 1871 to the present. It's far from clear that data from the late 19th century is reliable. It's even less likely that early data is useful for today. The world is a much different place, in terms of the economy, economic and political structures, market structures, investment structures, etc.
Even given that, we have fewer than three nonoverlapping 60 year periods since 1871. Predicting the future based on this many data points seems foolish. Having many independent data points is key for this sort of analysis. In particular, many independent, reliable and comparable data points.
They like an equity heavy portfolio. One reason is that they assume equities will have a high return: "To this end, we assume longterm average returns for equities going forward (about 6.6% real p.a.)".
I wouldn't rely on this.
Even given that, we have fewer than three nonoverlapping 60 year periods since 1871. Predicting the future based on this many data points seems foolish. Having many independent data points is key for this sort of analysis. In particular, many independent, reliable and comparable data points.
They like an equity heavy portfolio. One reason is that they assume equities will have a high return: "To this end, we assume longterm average returns for equities going forward (about 6.6% real p.a.)".
I wouldn't rely on this.

 Posts: 34
 Joined: Thu Feb 11, 2016 12:53 pm
Re: Study on SWR rates for Early Retirement
retiringtype wrote:At what age(s) is it considered early retirement? I'm 61. Is anything less than 65 "early?" Just curious.
For the sake of this thread, where the OP is looking at a 60 year time frame in retirement, that pushes early to sub 50 at least. More likely 40.
Re: Study on SWR rates for Early Retirement
I retired at age 59. Is this early retirement? Does the age of the spouse matter?
Re: Study on SWR rates for Early Retirement
The more I look into it, the less I like the idea of a safe withdrawal rate. I understand some people will say that the swr is just a number for planning purposes, and for that I guess a 3.5% SWR is rational with a 60 year time horizon. As a means of actually determining how much to withdraw every year in retirement I feel it falls well short of better withdrawal plans, such as Variable Percentage Withdrawal.
https://www.bogleheads.org/wiki/Variabl ... withdrawal
https://www.bogleheads.org/wiki/Variabl ... withdrawal
There are an army of people whose pay checks depend on convincing people to invest in ways that are against their self interest. This forum is the volunteer army that fights back!
Re: Study on SWR rates for Early Retirement
Quark wrote:They use data from 1871 to the present. It's far from clear that data from the late 19th century is reliable. It's even less likely that early data is useful for today. The world is a much different place, in terms of the economy, economic and political structures, market structures, investment structures, etc.
Even given that, we have fewer than three nonoverlapping 60 year periods since 1871. Predicting the future based on this many data points seems foolish. Having many independent data points is key for this sort of analysis. In particular, many independent, reliable and comparable data points.
They like an equity heavy portfolio. One reason is that they assume equities will have a high return: "To this end, we assume longterm average returns for equities going forward (about 6.6% real p.a.)".
I wouldn't rely on this.
Right.
There really is no way to estimate to any reasonable degree of accuracy what the SWR is for 60 years. The problem is that errors in assumptions grow exponentially. Even modest errors become huge out 30 years. Making claims for SWR out 60 years is just nonsensical.
If you "retire" at 35 and need to plan out 60 years, your only hope is to keep fixed expenses very low, and use a very flexible withdrawal rate.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.
Re: Study on SWR rates for Early Retirement
The best strategy is to pick something that seems sensible, then adjust based on changes in your portfolio and expenses. Realize that studies of the past have limited utility. Also realize that how to adjust is not going to be entirely clear.
Consider:
 Alan retires on 1/1/2016 with portfolio of $1,000,000. Using a 4% SWR, he can withdraw $40,000, inflation adjusted, for many years.
 The market drops 20% and Alan's portfolio is $800,000 on 1/1/2017. He can still withdraw $40,000/year
 Bob retires on 1/1/2017 with $800,000. Using a 4% SWR, he can withdraw $32,000, inflation adjusted, for many years.
Why should the results be different for Alan and Bob, given that both have the same portfolio on the same date? I've yet to see a good answer to this question.
Consider:
 Alan retires on 1/1/2016 with portfolio of $1,000,000. Using a 4% SWR, he can withdraw $40,000, inflation adjusted, for many years.
 The market drops 20% and Alan's portfolio is $800,000 on 1/1/2017. He can still withdraw $40,000/year
 Bob retires on 1/1/2017 with $800,000. Using a 4% SWR, he can withdraw $32,000, inflation adjusted, for many years.
Why should the results be different for Alan and Bob, given that both have the same portfolio on the same date? I've yet to see a good answer to this question.

 Posts: 635
 Joined: Tue Jul 31, 2007 9:36 pm
Re: Study on SWR rates for Early Retirement
Quark wrote:The best strategy is to pick something that seems sensible, then adjust based on changes in your portfolio and expenses. Realize that studies of the past have limited utility. Also realize that how to adjust is not going to be entirely clear.
Consider:
 Alan retires on 1/1/2016 with portfolio of $1,000,000. Using a 4% SWR, he can withdraw $40,000, inflation adjusted, for many years.
 The market drops 20% and Alan's portfolio is $800,000 on 1/1/2017. He can still withdraw $40,000/year
 Bob retires on 1/1/2017 with $800,000. Using a 4% SWR, he can withdraw $32,000, inflation adjusted, for many years.
Why should the results be different for Alan and Bob, given that both have the same portfolio on the same date? I've yet to see a good answer to this question.
Your question expects too much rigor from the model. The original Trinity study had a very simple construction and turned out to be very useful in *lowering* expectations during a bull market. Models are always wrong but can often be useful so finding a glaring paradox doesn't discredit the model.
Re: Study on SWR rates for Early Retirement
bradshaw1965 wrote:Quark wrote:The best strategy is to pick something that seems sensible, then adjust based on changes in your portfolio and expenses. Realize that studies of the past have limited utility. Also realize that how to adjust is not going to be entirely clear.
Consider:
 Alan retires on 1/1/2016 with portfolio of $1,000,000. Using a 4% SWR, he can withdraw $40,000, inflation adjusted, for many years.
 The market drops 20% and Alan's portfolio is $800,000 on 1/1/2017. He can still withdraw $40,000/year
 Bob retires on 1/1/2017 with $800,000. Using a 4% SWR, he can withdraw $32,000, inflation adjusted, for many years.
Why should the results be different for Alan and Bob, given that both have the same portfolio on the same date? I've yet to see a good answer to this question.
Your question expects too much rigor from the model. The original Trinity study had a very simple construction and turned out to be very useful in *lowering* expectations during a bull market. Models are always wrong but can often be useful so finding a glaring paradox doesn't discredit the model.
The point of the question is that the model is not the rigorous and is not to be followed blindly. There have been many posts here that take the model much too seriously.
Re: Study on SWR rates for Early Retirement
Quark, the table is merely a summary of the data. If you read the accompanying commentary, it does a very good job explaining the issues you bring up.
You are of course correct to mention that the political and economic landscape in the 1800's was very different than it is now. Of course the same can be said for the world 60 years from now. I agree with you that you need to start with a reasonable point and go from there. I would consider this table a reasonable starting point.
A key message is that if you are contemplating a retirement lasting greater than 30 years, the oft touted 4% initial withdrawal rate may not be a reasonable start.
You are of course correct to mention that the political and economic landscape in the 1800's was very different than it is now. Of course the same can be said for the world 60 years from now. I agree with you that you need to start with a reasonable point and go from there. I would consider this table a reasonable starting point.
A key message is that if you are contemplating a retirement lasting greater than 30 years, the oft touted 4% initial withdrawal rate may not be a reasonable start.
Last edited by czeckers on Tue Jan 03, 2017 2:05 pm, edited 1 time in total.
The Espresso portfolio: 

16% LCV, 16% SCV, 16% EM, 8% Int'l Value, 8% Int'l Sm, 8% US REIT, 8% Int'l REIT, 20% Interterm US Treas 

"A journey of a thousand miles begins with a single step."
Re: Study on SWR rates for Early Retirement
Early retirement in this case is a retirement expected to last longer than 30 years.
This is obviously affected by age at retirement as well as one's state of health.
This is obviously affected by age at retirement as well as one's state of health.
The Espresso portfolio: 

16% LCV, 16% SCV, 16% EM, 8% Int'l Value, 8% Int'l Sm, 8% US REIT, 8% Int'l REIT, 20% Interterm US Treas 

"A journey of a thousand miles begins with a single step."
Re: Study on SWR rates for Early Retirement
Rodc wrote:Quark wrote:They use data from 1871 to the present. It's far from clear that data from the late 19th century is reliable. It's even less likely that early data is useful for today. The world is a much different place, in terms of the economy, economic and political structures, market structures, investment structures, etc.
Even given that, we have fewer than three nonoverlapping 60 year periods since 1871. Predicting the future based on this many data points seems foolish. Having many independent data points is key for this sort of analysis. In particular, many independent, reliable and comparable data points.
They like an equity heavy portfolio. One reason is that they assume equities will have a high return: "To this end, we assume longterm average returns for equities going forward (about 6.6% real p.a.)".
I wouldn't rely on this.
Right.
There really is no way to estimate to any reasonable degree of accuracy what the SWR is for 60 years. The problem is that errors in assumptions grow exponentially. Even modest errors become huge out 30 years. Making claims for SWR out 60 years is just nonsensical.
If you "retire" at 35 and need to plan out 60 years, your only hope is to keep fixed expenses very low, and use a very flexible withdrawal rate.
It's your only hope if you box yourself into the plan and never change.
Retiring so early presumably means one has many skills. You could certainly use your human capital to bail yourself out of a problem by working part time or even full time. Or you could never fully "retire", however the internet retirement police decide to define that
Re: Study on SWR rates for Early Retirement
I will note that the original article uses a 100% equities portfolio for its calculations, which is not optimal based on portfolio theory, but particularly illsuited for a drawdown scenario. I'd be curious to see the results with a stillequity heavy 8020 stockbond portfolio, with rebalancing factored in to some degree/extent.
EDIT:
Oh, and one other note on these SWR studies: they define failure as any scenario where a portfolio drops to zero before the end of the time period (whether it be 30, 40, 50, or 60 years). So if 10 times out of 100 you run out of money, your success rate is 90%.
However, what frequently goes entirely UNEXPLORED by these studies is what amount you end up with in the 90% of SUCCESSFUL outcomes. Indeed, in some notinsignificant percentage of the 90% of successful scenarios, the portfolio grows to a number GREATER than it was when withdrawals began, meaning in those scenarios you could have afforded to draw out more money/increased your withdrawals to some degree during the entire period above the inflation adjustment.
All of which is to say that having some flexibility/employing a variable withdrawal method that keeps in mind returns/expenses dynamically is ideal, so that you don't outlive your money but also so that you don't underlive your money by not spending what you truly can/could have spent.
EDIT:
Oh, and one other note on these SWR studies: they define failure as any scenario where a portfolio drops to zero before the end of the time period (whether it be 30, 40, 50, or 60 years). So if 10 times out of 100 you run out of money, your success rate is 90%.
However, what frequently goes entirely UNEXPLORED by these studies is what amount you end up with in the 90% of SUCCESSFUL outcomes. Indeed, in some notinsignificant percentage of the 90% of successful scenarios, the portfolio grows to a number GREATER than it was when withdrawals began, meaning in those scenarios you could have afforded to draw out more money/increased your withdrawals to some degree during the entire period above the inflation adjustment.
All of which is to say that having some flexibility/employing a variable withdrawal method that keeps in mind returns/expenses dynamically is ideal, so that you don't outlive your money but also so that you don't underlive your money by not spending what you truly can/could have spent.
Re: Study on SWR rates for Early Retirement
Quark wrote:The best strategy is to pick something that seems sensible, then adjust based on changes in your portfolio and expenses. Realize that studies of the past have limited utility. Also realize that how to adjust is not going to be entirely clear.
Consider:
 Alan retires on 1/1/2016 with portfolio of $1,000,000. Using a 4% SWR, he can withdraw $40,000, inflation adjusted, for many years.
 The market drops 20% and Alan's portfolio is $800,000 on 1/1/2017. He can still withdraw $40,000/year
 Bob retires on 1/1/2017 with $800,000. Using a 4% SWR, he can withdraw $32,000, inflation adjusted, for many years.
Why should the results be different for Alan and Bob, given that both have the same portfolio on the same date? I've yet to see a good answer to this question.
If we were to assume that Alan and Bob had exactly the same portfolio (which seems to be the case in your example), Alan and Bob will have the same success rate using either $40K/year or $32K/year, inflation adjusted. This is because valuations and future performance does matter (as anyone who has seen the results of their portfolio testing for retirement dates in the midlate 60s can attest.)
In the case where Alan and Bob both start with $40K/year, Alan retired at a "peak" and Bob retired in a "valley", so Alan will have a lower chance of success with a lower SWR but Bob will have a higher chance of success with a higher SWR.
Re: Study on SWR rates for Early Retirement
MI_bogle wrote:Rodc wrote:Quark wrote:They use data from 1871 to the present. It's far from clear that data from the late 19th century is reliable. It's even less likely that early data is useful for today. The world is a much different place, in terms of the economy, economic and political structures, market structures, investment structures, etc.
Even given that, we have fewer than three nonoverlapping 60 year periods since 1871. Predicting the future based on this many data points seems foolish. Having many independent data points is key for this sort of analysis. In particular, many independent, reliable and comparable data points.
They like an equity heavy portfolio. One reason is that they assume equities will have a high return: "To this end, we assume longterm average returns for equities going forward (about 6.6% real p.a.)".
I wouldn't rely on this.
Right.
There really is no way to estimate to any reasonable degree of accuracy what the SWR is for 60 years. The problem is that errors in assumptions grow exponentially. Even modest errors become huge out 30 years. Making claims for SWR out 60 years is just nonsensical.
If you "retire" at 35 and need to plan out 60 years, your only hope is to keep fixed expenses very low, and use a very flexible withdrawal rate.
It's your only hope if you box yourself into the plan and never change.
Retiring so early presumably means one has many skills. You could certainly use your human capital to bail yourself out of a problem by working part time or even full time. Or you could never fully "retire", however the internet retirement police decide to define that
To a large extent that is what I said.
I almost mentioned the same point about human capital. But the problem is for many people getting a job with a meaningful paycheck in their 50s or 60s is very hard even with a reasonably impressive resume. Someone who has been a trust fund baby as many who "retire" at 35 are, or even if they were successful but are now out of work for 20 years, are really going to be in a bind if they need to generate an income. So in the end I decided that was not much of a backup plan. Might work of course.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.
Re: Study on SWR rates for Early Retirement
Quark wrote:
Consider:
 Alan retires on 1/1/2016 with portfolio of $1,000,000. Using a 4% SWR, he can withdraw $40,000, inflation adjusted, for many years.
 The market drops 20% and Alan's portfolio is $800,000 on 1/1/2017. He can still withdraw $40,000/year
 Bob retires on 1/1/2017 with $800,000. Using a 4% SWR, he can withdraw $32,000, inflation adjusted, for many years.
Why should the results be different for Alan and Bob, given that both have the same portfolio on the same date? I've yet to see a good answer to this question.
This points out the issue that starting valuations matter. In these types of studies, the failures occur as a result of people beginning retirement at the worst possible time such as right before a devastating bear market and thus represent the worst case scenario for planning purposes. Someone retiring immediately after a large market decline will be beginning with lower market valuations and thus will likely be able to sustain a higher than 4% SWR than the person beginning their retirement at a market top. Keep in mind that in most situations, a 4% SWR for a 30 year retirement will result in a surplus of money at death.
The Espresso portfolio: 

16% LCV, 16% SCV, 16% EM, 8% Int'l Value, 8% Int'l Sm, 8% US REIT, 8% Int'l REIT, 20% Interterm US Treas 

"A journey of a thousand miles begins with a single step."
Re: Study on SWR rates for Early Retirement
Those interested in the "meta" question, How to plan for (very) early retirement?, might want to check out this thread+poll I started a couple years ago: What SWR/expense multiple to use for early retirement plan?.
As every Boglehead knows, it's impossible to predict security prices, interest rates or the political climate of tomorrow; and it's even more impossible to predict anything 30 years out; and of course, it's most impossible to predict 60 years into the future. Recall what William Bernstein said in his Retirement Calculator from Hell, Part III: "any estimate of longterm financial success greater than about 80% is meaningless".
The linked study is indeed interesting, but as has been mentioned, fundamentally, there simply isn't enough data to build a robust model for very long retirement periods. Even if there was sufficient data, the meaning of that data changes over time; i.e., the world was a very different place 100 years ago, and it's likely to change again radically over the next 50+ years. I assume that's part of the reason why we haven't seen more studies of SWR rates for over 30 years (and also presumably because the "market" for such a study is fairly small).
Mr Money Mustache has been "selling" the idea of a 4% SWR ever since he started his blog. Though he doesn't appear willing to admit this outright, his idea of the "safe" in 4% SWR requires: dramatically lower expenses than most people ($25k/year for a family of three if he's the benchmark); some form of postretirement, nonportfolio income (his carpentry hobbyjob or $400k/year blog); 50% or so flexibility in the budget (look at his annual spending posts where he itemizes everything he could cut if he had to)...
I haven't looked at his work in a while, but I recall Wade Pfau used to be the king of pessimistic reports. IIRC, his work does support a 2% SWR for very longterm retirement planning.
Ultimately, I think you just have to educate yourself (well, that's true early retirement or not) and do a lot of selfreflection (also likely useful even if you're not considering early retirement). I think in many ways it's similar to the payoff mortgage early or invest more now discussion: in many cases, the math favors holding the mortgage, but how much value does one put in having no debt? Likewise, though there may be a lack of data (and therefore useful studies/models), clearly the math will always say that the lower the withdrawal rate, the safer the plan. But how much do you value early retirement versus some extra safety?
Very early retirement used to be top priority for me. It's still a goal, but age and various life experiences have made me focus more on financial security, rather than being able to retire. A big driver for this is health issues with one member of my family; I just filled out my FSA paperwork for 2016 and realized that my companyprovided insurance is paying over $400/month for meds. Of that, I only see a tiny copay. So despite that I track my finances very rigorously, this is one area where I've been blind to potential postretirement "surprise" expenses. And what if the issue gets worse?
At any rate, my planning method is similar to what is presented in Your Money or Your Life. And that is, track expenses versus passive (portfolio) income on a regular basis. I do this using GnuCash and a simple "dashboard" spreadsheet. Every now and then I use this data to generate a simple line chart, with time on the xasis and dollars on the yaxis. Two lines, one for expenses and one for portfolio income. Only when the income line has surpassed the expense line for at least a year will I start to think seriously about early retirement. Even then, given the personal issues I mentioned above, I might look to work parttime or see how I can make money from a "hobby job". Ideally I'd like enough income to cover my expenses, so I don't have to touch the portfolio for another 1020 years. Note the subtlety here: this implies that I'm looking to fund retirement exclusively by dividends (or bond payments); that is a deliberate move on my part in the interest of being conservative. I've been tracking finances this way since 2012. In those years, my dividend/interest income has been around 2.5% of portfolio value. Thus, I'm implicitly planning around that number as a SWR in retirement.
As every Boglehead knows, it's impossible to predict security prices, interest rates or the political climate of tomorrow; and it's even more impossible to predict anything 30 years out; and of course, it's most impossible to predict 60 years into the future. Recall what William Bernstein said in his Retirement Calculator from Hell, Part III: "any estimate of longterm financial success greater than about 80% is meaningless".
The linked study is indeed interesting, but as has been mentioned, fundamentally, there simply isn't enough data to build a robust model for very long retirement periods. Even if there was sufficient data, the meaning of that data changes over time; i.e., the world was a very different place 100 years ago, and it's likely to change again radically over the next 50+ years. I assume that's part of the reason why we haven't seen more studies of SWR rates for over 30 years (and also presumably because the "market" for such a study is fairly small).
Mr Money Mustache has been "selling" the idea of a 4% SWR ever since he started his blog. Though he doesn't appear willing to admit this outright, his idea of the "safe" in 4% SWR requires: dramatically lower expenses than most people ($25k/year for a family of three if he's the benchmark); some form of postretirement, nonportfolio income (his carpentry hobbyjob or $400k/year blog); 50% or so flexibility in the budget (look at his annual spending posts where he itemizes everything he could cut if he had to)...
I haven't looked at his work in a while, but I recall Wade Pfau used to be the king of pessimistic reports. IIRC, his work does support a 2% SWR for very longterm retirement planning.
Ultimately, I think you just have to educate yourself (well, that's true early retirement or not) and do a lot of selfreflection (also likely useful even if you're not considering early retirement). I think in many ways it's similar to the payoff mortgage early or invest more now discussion: in many cases, the math favors holding the mortgage, but how much value does one put in having no debt? Likewise, though there may be a lack of data (and therefore useful studies/models), clearly the math will always say that the lower the withdrawal rate, the safer the plan. But how much do you value early retirement versus some extra safety?
Very early retirement used to be top priority for me. It's still a goal, but age and various life experiences have made me focus more on financial security, rather than being able to retire. A big driver for this is health issues with one member of my family; I just filled out my FSA paperwork for 2016 and realized that my companyprovided insurance is paying over $400/month for meds. Of that, I only see a tiny copay. So despite that I track my finances very rigorously, this is one area where I've been blind to potential postretirement "surprise" expenses. And what if the issue gets worse?
At any rate, my planning method is similar to what is presented in Your Money or Your Life. And that is, track expenses versus passive (portfolio) income on a regular basis. I do this using GnuCash and a simple "dashboard" spreadsheet. Every now and then I use this data to generate a simple line chart, with time on the xasis and dollars on the yaxis. Two lines, one for expenses and one for portfolio income. Only when the income line has surpassed the expense line for at least a year will I start to think seriously about early retirement. Even then, given the personal issues I mentioned above, I might look to work parttime or see how I can make money from a "hobby job". Ideally I'd like enough income to cover my expenses, so I don't have to touch the portfolio for another 1020 years. Note the subtlety here: this implies that I'm looking to fund retirement exclusively by dividends (or bond payments); that is a deliberate move on my part in the interest of being conservative. I've been tracking finances this way since 2012. In those years, my dividend/interest income has been around 2.5% of portfolio value. Thus, I'm implicitly planning around that number as a SWR in retirement.
Re: Study on SWR rates for Early Retirement
Chadnudj wrote:I will note that the original article uses a 100% equities portfolio for its calculations.
The study looks at 100% 75% 50% 25% and 0% equities.
However, what frequently goes entirely UNEXPLORED by these studies is what amount you end up with in the 90% of SUCCESSFUL outcomes. Indeed, in some notinsignificant percentage of the 90% of successful scenarios, the portfolio grows to a number GREATER than it was when withdrawals began, meaning in those scenarios you could have afforded to draw out more money/increased your withdrawals to some degree during the entire period above the inflation adjustment.
This indeed is the case most of the time and the studies do look at that. Keep in mind that SWR studies are concerned about avoiding the worst case scenario of running out of money. Ending up with too much money is not viewed as a problem.
The Espresso portfolio: 

16% LCV, 16% SCV, 16% EM, 8% Int'l Value, 8% Int'l Sm, 8% US REIT, 8% Int'l REIT, 20% Interterm US Treas 

"A journey of a thousand miles begins with a single step."
Re: Study on SWR rates for Early Retirement
Suppose you are 40year Male old living in California. You will retire immediately and expect to live to 100, a 60 year retirement. You need inflationadjusted $1,000 per month, or $12,000 per year. How much will it cost to fund you retirement?
Based on TIPS prices as of 7/29/2016, can buy a 30Year TIPS ladder that pays $12,000 per year for $341,485. This will cover the first 30 years. See http://eyebonds.info/downloads/pages/TIPSLadder.html
At 2% inflation over 30 years, $1,000 compounds to $1,811.36.
Buy a deferred inflationadjusted annuity that that begins paying $1,811 for Life in 30 years.
Quote from immediateannuitues.com for 40 year old starting in 2047. Cost: $79,640 (without CPI adjustment)
I will guess 50% higher with CPI adjustment: $79,640 x 1.5 = $119,460
Total cost = $341,485 + $119,460 = $460945
12,000 / 460945 = 2.60% (38x)
This has 0% Stocks and 99+% chance of success.
Based on TIPS prices as of 7/29/2016, can buy a 30Year TIPS ladder that pays $12,000 per year for $341,485. This will cover the first 30 years. See http://eyebonds.info/downloads/pages/TIPSLadder.html
At 2% inflation over 30 years, $1,000 compounds to $1,811.36.
Buy a deferred inflationadjusted annuity that that begins paying $1,811 for Life in 30 years.
Quote from immediateannuitues.com for 40 year old starting in 2047. Cost: $79,640 (without CPI adjustment)
I will guess 50% higher with CPI adjustment: $79,640 x 1.5 = $119,460
Total cost = $341,485 + $119,460 = $460945
12,000 / 460945 = 2.60% (38x)
This has 0% Stocks and 99+% chance of success.
Gott mit uns.
Re: Study on SWR rates for Early Retirement
grayfox wrote:Suppose you are 40year Male old living in California. You will retire immediately and expect to live to 100, a 60 year retirement. You need inflationadjusted $1,000 per month, or $12,000 per year. How much will it cost to fund you retirement?
Based on TIPS prices as of 7/29/2016, can buy a 30Year TIPS ladder that pays $12,000 per year for $341,485. This will cover the first 30 years. See http://eyebonds.info/downloads/pages/TIPSLadder.html
At 2% inflation over 30 years, $1,000 compounds to $1,811.36.
Buy a deferred inflationadjusted annuity that that begins paying $1,811 for Life in 30 years.
Quote from immediateannuitues.com for 40 year old starting in 2047. Cost: $79,640 (without CPI adjustment)
I will guess 50% higher with CPI adjustment: $79,640 x 1.5 = $119,460
Total cost = $341,485 + $119,460 = $460945
12,000 / 460945 = 2.60% (38x)
This has 0% Stocks and 99+% chance of success.
Very interesting calculation. Thanks.
Not at all sure, but to get to 99% chance of success you may need to buy 2 or 3 such annuities from different companies since you are counting on getting full value out of the pension out to 60 years from now, but even so very interesting approach.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.
Re: Study on SWR rates for Early Retirement
czeckers wrote:Chadnudj wrote:I will note that the original article uses a 100% equities portfolio for its calculations.
The study looks at 100% 75% 50% 25% and 0% equities.However, what frequently goes entirely UNEXPLORED by these studies is what amount you end up with in the 90% of SUCCESSFUL outcomes. Indeed, in some notinsignificant percentage of the 90% of successful scenarios, the portfolio grows to a number GREATER than it was when withdrawals began, meaning in those scenarios you could have afforded to draw out more money/increased your withdrawals to some degree during the entire period above the inflation adjustment.
This indeed is the case most of the time and the studies do look at that. Keep in mind that SWR studies are concerned about avoiding the worst case scenario of running out of money. Ending up with too much money is not viewed as a problem.
I stand corrected. I think I got confused by following a link where he discussed why the 4% rule didn't work for those retiring near 2001, where he only ran the analysis on a 100% stock portfolio. But the point remains that the best portfolios in terms of longevity in draw down include some bonds (compare the 75% stock success rates to the 100%, and in pretty much each scenario below 4.25% you're better off with 75%).
As for the "worst case scenario," you are of course correct  best to plan for the worst, and hope for the best, obviously. But I think people go way overboard in panic ("4% is too risky! I have to do 3.25%!" "No, that's too much, too! I'd only do 2.75%!"), without realizing (a) you can improve your success rate by using variable withdrawal methods/reducing expenses, (b) the odds of you retiring on the WORST possible time in history and falling into that 1% failure (for 30 year retirement at 4% with 75% stocks) are much, much, MUCH worse than the odds of you ending up with more money than you started with (unsure of the percent, but it is significantly more than 1%, as I recall  maybe as high as 20% or more?).
EDIT: I found it. Per Kitces (here: https://www.kitces.com/blog/theratchet ... he4rule/):
In fact, by applying the 4% rule, over 2/3rds of the time the retiree finishes with more than double their wealth at the beginning of retirement, on top of a lifetime of (4% rule) spending! Half the time, wealth is nearly tripled by the end retirement, as retirees fail to spend their upside!
Seems like people are working a lot longer than they need to to protect against a 1% or so risk, when there is a 66% chance they end up with more than double their initial amount at retirement using the 4% rule.
EDIT 2:
Actually, it's only 25% of the time where people double their starting wealth, and 60% where they end up with at least 100% of their principle, once you factor in inflation:
Of course, in some of these situations, final wealth is augmented by decades of compounding inflation. Nonetheless, even on a real (inflationadjusted) basis, retirees finish with more than 100% of their inflationadjusted principal 60% of the time, and double their real wealth almost 1/4th of the time, even after supporting a lifetime of inflationadjusted spending at a 4% initial withdrawal rate!
Last edited by Chadnudj on Tue Jan 03, 2017 5:39 pm, edited 2 times in total.
Re: Study on SWR rates for Early Retirement
Whakamole wrote:Quark wrote:The best strategy is to pick something that seems sensible, then adjust based on changes in your portfolio and expenses. Realize that studies of the past have limited utility. Also realize that how to adjust is not going to be entirely clear.
Consider:
 Alan retires on 1/1/2016 with portfolio of $1,000,000. Using a 4% SWR, he can withdraw $40,000, inflation adjusted, for many years.
 The market drops 20% and Alan's portfolio is $800,000 on 1/1/2017. He can still withdraw $40,000/year
 Bob retires on 1/1/2017 with $800,000. Using a 4% SWR, he can withdraw $32,000, inflation adjusted, for many years.
Why should the results be different for Alan and Bob, given that both have the same portfolio on the same date? I've yet to see a good answer to this question.
If we were to assume that Alan and Bob had exactly the same portfolio (which seems to be the case in your example), Alan and Bob will have the same success rate using either $40K/year or $32K/year, inflation adjusted. This is because valuations and future performance does matter (as anyone who has seen the results of their portfolio testing for retirement dates in the midlate 60s can attest.)
In the case where Alan and Bob both start with $40K/year, Alan retired at a "peak" and Bob retired in a "valley", so Alan will have a lower chance of success with a lower SWR but Bob will have a higher chance of success with a higher SWR.
Forgot where I saw it, but someone pointed out that if Bob only needed $32k a year, he should have retired at least a year earlier (when his portfolio likely would have had nearly $1 million in it, too, like Alan's)
Re: Study on SWR rates for Early Retirement
Chadnudj wrote:czeckers wrote:Chadnudj wrote:I will note that the original article uses a 100% equities portfolio for its calculations.
The study looks at 100% 75% 50% 25% and 0% equities.However, what frequently goes entirely UNEXPLORED by these studies is what amount you end up with in the 90% of SUCCESSFUL outcomes. Indeed, in some notinsignificant percentage of the 90% of successful scenarios, the portfolio grows to a number GREATER than it was when withdrawals began, meaning in those scenarios you could have afforded to draw out more money/increased your withdrawals to some degree during the entire period above the inflation adjustment.
This indeed is the case most of the time and the studies do look at that. Keep in mind that SWR studies are concerned about avoiding the worst case scenario of running out of money. Ending up with too much money is not viewed as a problem.
I stand corrected. I think I got confused by following a link where he discussed why the 4% rule didn't work for those retiring near 2001, where he only ran the analysis on a 100% stock portfolio. But the point remains that the best portfolios in terms of longevity in draw down include some bonds (compare the 75% stock success rates to the 100%, and in pretty much each scenario below 4.25% you're better off with 75%).
As for the "worst case scenario," you are of course correct  best to plan for the worst, and hope for the best, obviously. But I think people go way overboard in panic ("4% is too risky! I have to do 3.25%!" "No, that's too much, too! I'd only do 2.75%!"), without realizing (a) you can improve your success rate by using variable withdrawal methods/reducing expenses, (b) the odds of you retiring on the WORST possible time in history and falling into that 1% failure (for 30 year retirement at 4% with 75% stocks) are much, much, MUCH worse than the odds of you ending up with more money than you started with (unsure of the percent, but it is significantly more than 1%, as I recall  maybe as high as 20% or more?).
EDIT: I found it. Per Kitces (here: https://www.kitces.com/blog/theratchet ... he4rule/):In fact, by applying the 4% rule, over 2/3rds of the time the retiree finishes with more than double their wealth at the beginning of retirement, on top of a lifetime of (4% rule) spending! Half the time, wealth is nearly tripled by the end retirement, as retirees fail to spend their upside!
Seems like people are working a lot longer than they need to to protect against a 1% or so risk, when there is a 66% chance they end up with more than double their initial amount at retirement using the 4% rule.
EDIT 2:
Actually, it's only 25% of the time where people double their starting wealth, and 60% where they end up with at least 100% of their principle, once you factor in inflation:Of course, in some of these situations, final wealth is augmented by decades of compounding inflation. Nonetheless, even on a real (inflationadjusted) basis, retirees finish with more than 100% of their inflationadjusted principal 60% of the time, and double their real wealth almost 1/4th of the time, even after supporting a lifetime of inflationadjusted spending at a 4% initial withdrawal rate!
+1... I try to express this same sentiment every time we start getting into a competition over how low of a starting withdrawal rate one can suggest.
Can we all agree that suggesting withdrawal rates of under 3% in the name of prudence is probably misguided regardless of the proposed time frame?
Re: Study on SWR rates for Early Retirement
bigred77 wrote:Can we all agree that suggesting withdrawal rates of under 3% in the name of prudence is probably misguided regardless of the proposed time frame?
+1
Re: Study on SWR rates for Early Retirement
People seek certainty about the distant future, so some try to supply a semblance of it. Of course, if you spend to fit the worst case, then you have underspent for the other 95% of the cases. That is some expensive insurance.
If you accept the risk of variable retirement income by having most of your necessary expenses covered by more steady income sources (delayed SS with its COLA, annuity ladders, pension), that leaves most of the variable income for discretionary spending. Why not expect to adapt to the market returns instead of trying to guess within a tenth of a percent, what the future average returns will be?
Seeking that certainty just seems like the wrong assumption from which to start, except for the public demand for it. It is just a subcase of the same illusion in the active vs. passive management competition for investment assets.
If you accept the risk of variable retirement income by having most of your necessary expenses covered by more steady income sources (delayed SS with its COLA, annuity ladders, pension), that leaves most of the variable income for discretionary spending. Why not expect to adapt to the market returns instead of trying to guess within a tenth of a percent, what the future average returns will be?
Seeking that certainty just seems like the wrong assumption from which to start, except for the public demand for it. It is just a subcase of the same illusion in the active vs. passive management competition for investment assets.
Re: Study on SWR rates for Early Retirement
There was another thread about calculating the Funding Ratio FR = PV(Assets) / PV(Liabilities)
Fully funded would have a Funding Ratio of 1.00.
I believe this is what many Financial Economists would look at to determine if a Pension Plan has enough assets to meet its obligations.
It would probably be prudent to build in a Margin for Error, due to uncertainties. So may be FR > 1.20
One of the posters linked to a spreadsheet which calculates the Funding Ratio, which makes it easy.
For the same 40year old guy that requires inflationadjusted $12,000 per year until age 100. Assuming no SS or Pension.
Backtesting showed that, with 75% Stocks, 3.25% (30.77x) had 100% Success Rate. That would imply
Assets = 30.77 x 12,000 = $369,231
Funding Ratio = 369,213/ 539,460 = 0.68
That early retirement is woefully underfunded.
Fully funded would have a Funding Ratio of 1.00.
I believe this is what many Financial Economists would look at to determine if a Pension Plan has enough assets to meet its obligations.
It would probably be prudent to build in a Margin for Error, due to uncertainties. So may be FR > 1.20
One of the posters linked to a spreadsheet which calculates the Funding Ratio, which makes it easy.
For the same 40year old guy that requires inflationadjusted $12,000 per year until age 100. Assuming no SS or Pension.
Backtesting showed that, with 75% Stocks, 3.25% (30.77x) had 100% Success Rate. That would imply
Assets = 30.77 x 12,000 = $369,231
Funding Ratio = 369,213/ 539,460 = 0.68
That early retirement is woefully underfunded.
Gott mit uns.
Re: Study on SWR rates for Early Retirement
grayfox wrote:There was another thread about calculating the Funding Ratio FR = PV(Assets) / PV(Liabilities)
Fully funded would have a Funding Ratio of 1.00.
I believe this is what many Financial Economists would look at to determine if a Pension Plan has enough assets to meet its obligations.
It would probably be prudent to build in a Margin for Error, due to uncertainties. So may be FR > 1.20
One of the posters linked to a spreadsheet which calculates the Funding Ratio, which makes it easy.
For the same 40year old guy that requires inflationadjusted $12,000 per year until age 100. Assuming no SS or Pension.
Backtesting showed that, with 75% Stocks, 3.25% (30.77x) had 100% Success Rate. That would imply
Assets = 30.77 x 12,000 = $369,231
Funding Ratio = 369,213/ 539,460 = 0.68
That early retirement is woefully underfunded.
Unfortunately the funding ratio is based on funding retirement with a TIPS ladder which is going to be very tough with a 60 year horizon.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.
Re: Study on SWR rates for Early Retirement
Rodc wrote:Unfortunately the funding ratio is based on funding retirement with a TIPS ladder which is going to be very tough with a 60 year horizon.
Right, that is the problem with calculating Present Values. Which discount rate should be used? The spreadsheet uses a 1% discount rate, which is about the 30year TIPS rate. Actually, today the real yield = 0.94%.
Gott mit uns.
Re: Study on SWR rates for Early Retirement
grayfox wrote:Rodc wrote:Unfortunately the funding ratio is based on funding retirement with a TIPS ladder which is going to be very tough with a 60 year horizon.
Right, that is the problem with calculating Present Values. Which discount rate should be used? The spreadsheet uses a 1% discount rate, which is about the 30year TIPS rate. Actually, today the real yield = 0.94%.
I played with this some. One can up the discount rate to say assume a higher return based on holding stocks and bonds. But the PV calculation completely misses the "poor sequence of returns" problem and quickly becomes overly optimistic. So while interesting, it is unclear how valuable the funding ratio is for people who really are not going to use a TIPS ladder to fund their retirement.
Your TIPS (liability matching portfolio) approach coupled with an annuity is likely a better approach.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.
Re: Study on SWR rates for Early Retirement
grayfox wrote:Suppose you are 40year Male old living in California. You will retire immediately and expect to live to 100, a 60 year retirement. You need inflationadjusted $1,000 per month, or $12,000 per year. How much will it cost to fund you retirement?
Based on TIPS prices as of 7/29/2016, can buy a 30Year TIPS ladder that pays $12,000 per year for $341,485. This will cover the first 30 years. See http://eyebonds.info/downloads/pages/TIPSLadder.html
At 2% inflation over 30 years, $1,000 compounds to $1,811.36.
Buy a deferred inflationadjusted annuity that that begins paying $1,811 for Life in 30 years.
Quote from immediateannuitues.com for 40 year old starting in 2047. Cost: $79,640 (without CPI adjustment)
I will guess 50% higher with CPI adjustment: $79,640 x 1.5 = $119,460
Total cost = $341,485 + $119,460 = $460945
12,000 / 460945 = 2.60% (38x)
This has 0% Stocks and 99+% chance of success.
Is it actually possible to buy a deferred inflationadjusted annuity? For that matter, is it actually possible to buy any inflation adjusted annuity? There used to be three companies that sold immediate inflation adjusted annuities, but I can no longer find any. The closest I can find are some that adjust by 2%/year.
The odds are high that something unexpected happens over a 60 year span.
Re: Study on SWR rates for Early Retirement
Rodc wrote:grayfox wrote:Rodc wrote:Unfortunately the funding ratio is based on funding retirement with a TIPS ladder which is going to be very tough with a 60 year horizon.
Right, that is the problem with calculating Present Values. Which discount rate should be used? The spreadsheet uses a 1% discount rate, which is about the 30year TIPS rate. Actually, today the real yield = 0.94%.
I played with this some. One can up the discount rate to say assume a higher return based on holding stocks and bonds. But the PV calculation completely misses the "poor sequence of returns" problem and quickly becomes overly optimistic. So while interesting, it is unclear how valuable the funding ratio is for people who really are not going to use a TIPS ladder to fund their retirement.
Your TIPS (liability matching portfolio) approach coupled with an annuity is likely a better approach.
If you're going to use a TIPS ladder, why not just price a TIPS ladder rather than calculate a funding ratio? If you're going to also use an annuity, you can price that too.
Re: Study on SWR rates for Early Retirement
Quark wrote:Rodc wrote:grayfox wrote:Rodc wrote:Unfortunately the funding ratio is based on funding retirement with a TIPS ladder which is going to be very tough with a 60 year horizon.
Right, that is the problem with calculating Present Values. Which discount rate should be used? The spreadsheet uses a 1% discount rate, which is about the 30year TIPS rate. Actually, today the real yield = 0.94%.
I played with this some. One can up the discount rate to say assume a higher return based on holding stocks and bonds. But the PV calculation completely misses the "poor sequence of returns" problem and quickly becomes overly optimistic. So while interesting, it is unclear how valuable the funding ratio is for people who really are not going to use a TIPS ladder to fund their retirement.
Your TIPS (liability matching portfolio) approach coupled with an annuity is likely a better approach.
If you're going to use a TIPS ladder, why not just price a TIPS ladder rather than calculate a funding ratio? If you're going to also use an annuity, you can price that too.
As BobK laid out the idea is you do price a TIPS ladder and then just look at how much you have now and see if you have the money (or rather what fraction of the money you have). The thing is what you have now might in part be your 401K, but might also be some future nominal pension, etc. You can use the PV calculation to aggregate your various sources of future income. Certainly not the only approach. Not sure why he uses a ratio vs difference. In the end I care more about I need to save an additional $X to get the income I want rather than I have 72% of what I need  though of course I can easily translate one to the other.
In the end though likely not a very good estimation technique for folks holding stocks and bonds and facing a long retirement.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.
Re: Study on SWR rates for Early Retirement
Rodc wrote:Quark wrote:Rodc wrote:grayfox wrote:Rodc wrote:Unfortunately the funding ratio is based on funding retirement with a TIPS ladder which is going to be very tough with a 60 year horizon.
Right, that is the problem with calculating Present Values. Which discount rate should be used? The spreadsheet uses a 1% discount rate, which is about the 30year TIPS rate. Actually, today the real yield = 0.94%.
I played with this some. One can up the discount rate to say assume a higher return based on holding stocks and bonds. But the PV calculation completely misses the "poor sequence of returns" problem and quickly becomes overly optimistic. So while interesting, it is unclear how valuable the funding ratio is for people who really are not going to use a TIPS ladder to fund their retirement.
Your TIPS (liability matching portfolio) approach coupled with an annuity is likely a better approach.
If you're going to use a TIPS ladder, why not just price a TIPS ladder rather than calculate a funding ratio? If you're going to also use an annuity, you can price that too.
As BobK laid out the idea is you do price a TIPS ladder and then just look at how much you have now and see if you have the money (or rather what fraction of the money you have). The thing is what you have now might in part be your 401K, but might also be some future nominal pension, etc. You can use the PV calculation to aggregate your various sources of future income. Certainly not the only approach. Not sure why he uses a ratio vs difference. In the end I care more about I need to save an additional $X to get the income I want rather than I have 72% of what I need  though of course I can easily translate one to the other.
In the end though likely not a very good estimation technique for folks holding stocks and bonds and facing a long retirement.
A simple PV calculation is not a very good estimation technique for a TIPS ladder, given a rising yield curve (although it should be in the general ballpark). It's certainly not a very good estimation technique for an annuity or for stocks and bonds. We also learn that to be safely funded you should add 520% to the calculated funding ratio.
Using actual prices for guaranteed safe assets (or the like) has the virtue of accuracy. If you're going to fund your retirement with stocks or other risky assets (as seems the case for the vast majority needing more than Social Security or pensions), any technique (funding ratio, SWR, etc.) is going to be an approximation. If you're going to fund your retirement with Social Security, the calculations should be very easy.
A moderate SWR with the knowledge that it's going to require adjustment seems as good as anything, unless you're going to fund with guaranteed assets. Moderate likely means lower than historical. How much lower seems a matter of taste.
Re: Study on SWR rates for Early Retirement
Quark wrote:Rodc wrote:Quark wrote:Rodc wrote:grayfox wrote:Right, that is the problem with calculating Present Values. Which discount rate should be used? The spreadsheet uses a 1% discount rate, which is about the 30year TIPS rate. Actually, today the real yield = 0.94%.
I played with this some. One can up the discount rate to say assume a higher return based on holding stocks and bonds. But the PV calculation completely misses the "poor sequence of returns" problem and quickly becomes overly optimistic. So while interesting, it is unclear how valuable the funding ratio is for people who really are not going to use a TIPS ladder to fund their retirement.
Your TIPS (liability matching portfolio) approach coupled with an annuity is likely a better approach.
If you're going to use a TIPS ladder, why not just price a TIPS ladder rather than calculate a funding ratio? If you're going to also use an annuity, you can price that too.
As BobK laid out the idea is you do price a TIPS ladder and then just look at how much you have now and see if you have the money (or rather what fraction of the money you have). The thing is what you have now might in part be your 401K, but might also be some future nominal pension, etc. You can use the PV calculation to aggregate your various sources of future income. Certainly not the only approach. Not sure why he uses a ratio vs difference. In the end I care more about I need to save an additional $X to get the income I want rather than I have 72% of what I need  though of course I can easily translate one to the other.
In the end though likely not a very good estimation technique for folks holding stocks and bonds and facing a long retirement.
A simple PV calculation is not a very good estimation technique for a TIPS ladder, given a rising yield curve (although it should be in the general ballpark). It's certainly not a very good estimation technique for an annuity or for stocks and bonds. We also learn that to be safely funded you should add 520% to the calculated funding ratio.
Using actual prices for guaranteed safe assets (or the like) has the virtue of accuracy. If you're going to fund your retirement with stocks or other risky assets (as seems the case for the vast majority needing more than Social Security or pensions), any technique (funding ratio, SWR, etc.) is going to be an approximation. If you're going to fund your retirement with Social Security, the calculations should be very easy.
A moderate SWR with the knowledge that it's going to require adjustment seems as good as anything, unless you're going to fund with guaranteed assets. Moderate likely means lower than historical. How much lower seems a matter of taste.
I agree with all that.
Assuming a constant real yield makes this only an estimate. Using the 10year TIPS (or something in that ballpark) yield for a 20 or 30 year ladder would be better than using the actual long term yield (too high in the early years and too low in the out years). Of course if one were desiring a TIPS ladder today you would not estimate anything, you would just use what is available today at today's prices. I have not taken the time to quantify an estimate of the error due to a rising yield curve, but I expect all in all it is not a bad TIPS ladder estimate, but yes it is just an estimate. And if one is thinking about buying a TIPS ladder some years down the road, even using today's yield curve will have errors, possibly of the same magnitude as just assuming some ballpark constant yield.
If one is thinking of retiring soon, and wants certainty, simply pricing an annuity is the best thing to do. If making a plan out many years every method is something of a crap shoot because the price is subject to nontrivial changes.
Agree with the last as well and is my primary method as I approach retirement. One thing I don't like about the funding ratio is the dependence on picking a date of death. If you are older and so not looking out more than 20 or maybe 30 years it seems to give sensible answers, but in the context of this thread, where the horizon is 3060 years the assumed 1% growth TIPS ladder ("like") funding model breaks down.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

 Posts: 454
 Joined: Tue May 21, 2013 8:49 pm
Re: Study on SWR rates for Early Retirement
If you're looking at SWR for the very young, then you might look at "perpetual withdrawal rates"
I found this paper by Otar to be very interesting: http://retirementoptimizer.com/Whitepapers/PerpDist.pdf
For those looking at superearly retirement, I think it's worth thinking about this as an "escape velocity" problem. You want enough, and to be withdrawing low enough, that the nest egg is likely to grow. As it grows faster than you're drawing on it, you build up a bigger safety margin to handle [bad stuff happening] in the future, where bad stuff could be a crash, changes to tax law, or unexpectedly high increase in health care or other expenses etc
I found this paper by Otar to be very interesting: http://retirementoptimizer.com/Whitepapers/PerpDist.pdf
For those looking at superearly retirement, I think it's worth thinking about this as an "escape velocity" problem. You want enough, and to be withdrawing low enough, that the nest egg is likely to grow. As it grows faster than you're drawing on it, you build up a bigger safety margin to handle [bad stuff happening] in the future, where bad stuff could be a crash, changes to tax law, or unexpectedly high increase in health care or other expenses etc
Re: Study on SWR rates for Early Retirement
Quark wrote:Is it actually possible to buy a deferred inflationadjusted annuity? For that matter, is it actually possible to buy any inflation adjusted annuity? There used to be three companies that sold immediate inflation adjusted annuities, but I can no longer find any. The closest I can find are some that adjust by 2%/year.
The odds are high that something unexpected happens over a 60 year span.
Good question!
You can still get a quote for CPIadjusted annuity through Vanguard, that pays immediate and deferred. However, it appears that fewer insurance companies now are offering it. Last March I checked and there were multiple companies including AIG who gave quotes. A couple of days ago I checked and only one company.
Also, the price has also gone up compared to noninflation adjusted annuity. Here is how much more you had had to pay for CPIadjusted annuity over no inflation adjustment, starting in 0, 5, 10, 15 years.
Code: Select all
0 5 10 15
Mar2016 39% 34% 28% 30%
Dec2016 50% 40% 31% 24%
Gott mit uns.

 Posts: 1026
 Joined: Tue Feb 23, 2016 4:24 am
Re: Study on SWR rates for Early Retirement
WageSlave wrote:Those interested in the "meta" question, How to plan for (very) early retirement?, might want to check out this thread+poll I started a couple years ago: What SWR/expense multiple to use for early retirement plan?.
As every Boglehead knows, it's impossible to predict security prices, interest rates or the political climate of tomorrow; and it's even more impossible to predict anything 30 years out; and of course, it's most impossible to predict 60 years into the future. Recall what William Bernstein said in his Retirement Calculator from Hell, Part III: "any estimate of longterm financial success greater than about 80% is meaningless".
The linked study is indeed interesting, but as has been mentioned, fundamentally, there simply isn't enough data to build a robust model for very long retirement periods. Even if there was sufficient data, the meaning of that data changes over time; i.e., the world was a very different place 100 years ago, and it's likely to change again radically over the next 50+ years. I assume that's part of the reason why we haven't seen more studies of SWR rates for over 30 years (and also presumably because the "market" for such a study is fairly small).
Mr Money Mustache has been "selling" the idea of a 4% SWR ever since he started his blog. Though he doesn't appear willing to admit this outright, his idea of the "safe" in 4% SWR requires: dramatically lower expenses than most people ($25k/year for a family of three if he's the benchmark); some form of postretirement, nonportfolio income (his carpentry hobbyjob or $400k/year blog); 50% or so flexibility in the budget (look at his annual spending posts where he itemizes everything he could cut if he had to)...
I haven't looked at his work in a while, but I recall Wade Pfau used to be the king of pessimistic reports. IIRC, his work does support a 2% SWR for very longterm retirement planning.
Ultimately, I think you just have to educate yourself (well, that's true early retirement or not) and do a lot of selfreflection (also likely useful even if you're not considering early retirement). I think in many ways it's similar to the payoff mortgage early or invest more now discussion: in many cases, the math favors holding the mortgage, but how much value does one put in having no debt? Likewise, though there may be a lack of data (and therefore useful studies/models), clearly the math will always say that the lower the withdrawal rate, the safer the plan. But how much do you value early retirement versus some extra safety?
Very early retirement used to be top priority for me. It's still a goal, but age and various life experiences have made me focus more on financial security, rather than being able to retire. A big driver for this is health issues with one member of my family; I just filled out my FSA paperwork for 2016 and realized that my companyprovided insurance is paying over $400/month for meds. Of that, I only see a tiny copay. So despite that I track my finances very rigorously, this is one area where I've been blind to potential postretirement "surprise" expenses. And what if the issue gets worse?
At any rate, my planning method is similar to what is presented in Your Money or Your Life. And that is, track expenses versus passive (portfolio) income on a regular basis. I do this using GnuCash and a simple "dashboard" spreadsheet. Every now and then I use this data to generate a simple line chart, with time on the xasis and dollars on the yaxis. Two lines, one for expenses and one for portfolio income. Only when the income line has surpassed the expense line for at least a year will I start to think seriously about early retirement. Even then, given the personal issues I mentioned above, I might look to work parttime or see how I can make money from a "hobby job". Ideally I'd like enough income to cover my expenses, so I don't have to touch the portfolio for another 1020 years. Note the subtlety here: this implies that I'm looking to fund retirement exclusively by dividends (or bond payments); that is a deliberate move on my part in the interest of being conservative. I've been tracking finances this way since 2012. In those years, my dividend/interest income has been around 2.5% of portfolio value. Thus, I'm implicitly planning around that number as a SWR in retirement.
Absolutely agree that saying 95 percent or more at 60 years is crazy
For one 60 year period
https://en.m.wikipedia.org/wiki/The_Man ... (TV_series)
For 2 you are back in reconstruction era
A little after
https://en.m.wikipedia.org/wiki/Gone_wi ... Wind_(film)
You cannot make reasonable plans for that long
Who knows what the next 60 years will bring?
https://en.m.wikipedia.org/wiki/Terminator_(franchise)
G.E. Box "All models are wrong, but some are useful."
Re: Study on SWR rates for Early Retirement
When looking at 3060 year time frames, an annuity without an inflation adjustment would make me very nervous.
The Espresso portfolio: 

16% LCV, 16% SCV, 16% EM, 8% Int'l Value, 8% Int'l Sm, 8% US REIT, 8% Int'l REIT, 20% Interterm US Treas 

"A journey of a thousand miles begins with a single step."
Re: Study on SWR rates for Early Retirement
Chadnudj wrote:
Seems like people are working a lot longer than they need to to protect against a 1% or so risk, when there is a 66% chance they end up with more than double their initial amount at retirement using the 4% rule.
I 100% agree with you. If you consider these SWR rates based on market history including the 90% drop and subsequent Great Depression, even at current valuations, it's hard to envision that we are starting at a worse point than that. There is a indeed a real sacrifice in working much longer or constraining spending much more than necessary when you are looking at SWR rates significantly below the value(s) shown in the table for your given situation.
The other piece of data I wish we had was to see what effect international diversification has on the results. All the studies so far (Bengen, Trinity, and this) only use US market data. There are definitely times when international diversification would have helped to mitigate portfolio losses such as during the oil embargo.
Unfortunately, reliable historical data on international markets is even harder to come by than for US data. (And I agree that there are a lot of caveats with looking at US data prior to 1917.)
K
The Espresso portfolio: 

16% LCV, 16% SCV, 16% EM, 8% Int'l Value, 8% Int'l Sm, 8% US REIT, 8% Int'l REIT, 20% Interterm US Treas 

"A journey of a thousand miles begins with a single step."
Re: Study on SWR rates for Early Retirement
random_walker_77 wrote:If you're looking at SWR for the very young, then you might look at "perpetual withdrawal rates"
I found this paper by Otar to be very interesting: http://retirementoptimizer.com/Whitepapers/PerpDist.pdf
For those looking at superearly retirement, I think it's worth thinking about this as an "escape velocity" problem. You want enough, and to be withdrawing low enough, that the nest egg is likely to grow. As it grows faster than you're drawing on it, you build up a bigger safety margin to handle [bad stuff happening] in the future, where bad stuff could be a crash, changes to tax law, or unexpectedly high increase in health care or other expenses etc
I gave a quick look at that paper. On page 2, Otar says
If you only want a fixed dollar amount of distributions, then the perpetual distribution rate is 2.8% per year, calculated as a percentage of the initial asset value.
If you want a fixed dollar amount of distributions that are indexed to CPI, then the starting perpetual distribution rate is 2.3% percent, calculated as a percentage of the initial asset value.
So he says that Trinitystyle perpetual withdrawal rate is 2.3%. That is much lower than the 3.25% that one would get from the Table above. In fact, it is close to the answer you get from for a Fully Funded Pension.
Back in 2011, I looked into the question of Perpetual Withdrawal. I wanted to determine a perpetual withdrawal rate from 100% stocks from basic economic principles without relying on backtesting to find the answer. For the S&P500, I came up with annual withdrawal rate = 0.80*E10/P. (The 0.80 was an arbitrary margin of safety factor to account for various uncertainties.) Another Boglehad Poster, CJKing, came up with almost the same answer using historical data.
Today P/E10 = 28.03. E10/P = 3.5676 and 0.80*E10/P = 2.854%
E.g. If your portfolio value at the end of the 2016 is $1,000,000 and P/E10 = 28.03. Withdrawer $28,540 on January 1, 2017.
Do the calculation and withdrawal every year.
Note that this gives a Variable withdrawal amount, not a constant inflationadjusted amount.
Last edited by grayfox on Wed Jan 04, 2017 12:08 pm, edited 1 time in total.
Gott mit uns.
Re: Study on SWR rates for Early Retirement
I think of perpetual withdrawal rates when I think of endowments. 60 years is a long time, but it is definitely finite. Add to that that there is a probability that one will die earlier than expected and, for me, the extra sacrifices to fund a retirement at 2.3% rather than 3.25 or 3.5% simply aren't worth it. It's the difference between having 30x and 43x annual expenses saved up at the start of retirement.
If you are very wealthy and have no concerns about not having enough money, and are rather concerned with leaving a legacy to your heirs, then that's a different story.
K
If you are very wealthy and have no concerns about not having enough money, and are rather concerned with leaving a legacy to your heirs, then that's a different story.
K
The Espresso portfolio: 

16% LCV, 16% SCV, 16% EM, 8% Int'l Value, 8% Int'l Sm, 8% US REIT, 8% Int'l REIT, 20% Interterm US Treas 

"A journey of a thousand miles begins with a single step."
Re: Study on SWR rates for Early Retirement
Thanks for the link, OP. Very interesting!
+1. This is my approach as well, and I recently wrote an explanation of how perpetual withdrawal rates work and why they're a good option for very early retirees: Perpetual Withdrawal Rates Are The Runway To A Long Retirement
Try this: https://portfoliocharts.com/portfolio/withdrawalrates/
random_walker_77 wrote:If you're looking at SWR for the very young, then you might look at "perpetual withdrawal rates"
+1. This is my approach as well, and I recently wrote an explanation of how perpetual withdrawal rates work and why they're a good option for very early retirees: Perpetual Withdrawal Rates Are The Runway To A Long Retirement
czeckers wrote:The other piece of data I wish we had was to see what effect international diversification has on the results. All the studies so far (Bengen, Trinity, and this) only use US market data. There are definitely times when international diversification would have helped to mitigate portfolio losses such as during the oil embargo.
Try this: https://portfoliocharts.com/portfolio/withdrawalrates/
Re: Study on SWR rates for Early Retirement
czeckers wrote:When looking at 3060 year time frames, an annuity without an inflation adjustment would make me very nervous.
I'd agree except you are definitely paying for that inflation adjustment. (Either that or the insurance company is discounting the possibility of high inflation, in which case you may end up in bad shape  see how all the LTC insurance underwriters are doing nowadays.)