Will we be OK at 3% WR?

Have a question about your personal investments? No matter how simple or complex, you can ask it here.
Rob1
Posts: 153
Joined: Thu May 04, 2017 2:57 pm

Re: Will we be OK at 3% WR?

Post by Rob1 » Fri Nov 08, 2019 3:33 pm

Garfieldthecat wrote:
Wed Nov 06, 2019 7:56 am
I can't find the webpage, but I've seen it linked in the forum a couple of times, where it basically says a success rate of 95% is fine, because you can't predict an asteroid crashing into us, or WWIII starting, or some other huge disaster.
"But history teaches us that depriving ourselves to boost our 40-year success probability much beyond 80% is a fool’s errand, since all you are doing is increasing the probability of failure for political, economic, and military reasons relative to the failure of banal financial planning."

More context in Bernstein‘s article here:
The Retirement Calculator from Hell, Part III
http://www.efficientfrontier.com/ef/901/hell3.htm

Thesaints
Posts: 2905
Joined: Tue Jun 20, 2017 12:25 am

Re: Will we be OK at 3% WR?

Post by Thesaints » Fri Nov 08, 2019 3:46 pm

CFM300 wrote:
Thu Nov 07, 2019 9:50 pm
You might find this link interesting and useful:

https://portfoliocharts.com/portfolio/withdrawal-rates/

Not only does it explain (albeit briefly) the back-testing methodologies employed in previous studies, it has a fantastic calculator that allows you to see both the safe withdrawal rate (which means "did not run out of money in worst case") and the perpetual withdrawal rate (which means "sustained initial principal in worst case").

The default setting is 60/40 (TSM / intermediate-term bonds). It shows a safe withdrawal rate of 4.5% and a perpetual withdrawal rate of 3.5% over 30 years.
I'm sorry. There must be some sort of disconnect.

By "did not run out in the worst case" they mean the worst case "based on real-life sequence of returns". "Perpetual withdrawal rate", changes the constraint from "still having money after 30 years" (which can mean having as little as $1) to "still having the initial capital after 30 years" (which, effectively means "in perpetuity" since one can start another 30 years). This also is based on "based on real-life sequence of returns".

The Trinity Study, which many like to cite here, had a yet different (and looser) constraint "still having money after 30 years in at least x% of cases". Obviously, the Trinity Study cannot be based on the "worst possible case" historically based or not. It is instead based on all historical returns series, good and bad.

Once we accept the above, my objection is that past history average stock returns has been 10% and bond return has been 7%. This is obviously true for the 90 years of recorded history, for 40-year long periods, and for most 30-year long ones, which are relevant for retirement planning purposes. It is not true for 10-years long periods; there are many with quite different average returns, but those are not too relevant for retirement purposes.

Now, drum-roll, things are different for developed markets. They have been different for Japan for the last 30 years. They will be different for Europe and the US, until monetary policies (i.e. loose money) change and there is no reason why they should change. We can discuss this separately.
Present and foreseable monetary policies tell us that we can only dream of 10% average returns for stocks and, above all, bond returns will be close to zero real.
That means average balanced portfolio returns won't be able to sustain past "safe" withdrawal rates with the same chance of success for two reasons: lower average returns and highe portfolio volatility (stock returns will be lower, but not their volatility !).
Continue to plan for 4% SWR being oblivious of this fundamental change in financial markets and your chances of disappointment will be higher.

User avatar
FiveK
Posts: 7741
Joined: Sun Mar 16, 2014 2:43 pm

Re: Will we be OK at 3% WR?

Post by FiveK » Fri Nov 08, 2019 3:51 pm

Thesaints wrote:
Fri Nov 08, 2019 3:46 pm
...my objection is that past history average stock returns has been 10% and bond return has been 7%.
What do you think the returns were in the worst 30 year period that was successful?

Thesaints
Posts: 2905
Joined: Tue Jun 20, 2017 12:25 am

Re: Will we be OK at 3% WR?

Post by Thesaints » Fri Nov 08, 2019 4:04 pm

Table at the bottom contains all inflation-adjusted numbers

https://fourpillarfreedom.com/heres-how ... ince-1928/

The lowest (of all!) 30-year period is 4.3%. Japanese last 30-year return ? Less than zero.

User avatar
FiveK
Posts: 7741
Joined: Sun Mar 16, 2014 2:43 pm

Re: Will we be OK at 3% WR?

Post by FiveK » Fri Nov 08, 2019 4:13 pm

Thesaints wrote:
Fri Nov 08, 2019 4:04 pm
Table at the bottom contains all inflation-adjusted numbers

https://fourpillarfreedom.com/heres-how ... ince-1928/

The lowest (of all!) 30-year period is 4.3%. Japanese last 30-year return ? Less than zero.
Are you suggesting that one should use a negative return on investments for planning purposes?

JackoC
Posts: 1005
Joined: Sun Aug 12, 2018 11:14 am

Re: Will we be OK at 3% WR?

Post by JackoC » Fri Nov 08, 2019 4:24 pm

Thesaints wrote:
Fri Nov 08, 2019 3:46 pm
CFM300 wrote:
Thu Nov 07, 2019 9:50 pm
You might find this link interesting and useful:

https://portfoliocharts.com/portfolio/withdrawal-rates/

Not only does it explain (albeit briefly) the back-testing methodologies employed in previous studies, it has a fantastic calculator that allows you to see both the safe withdrawal rate (which means "did not run out of money in worst case") and the perpetual withdrawal rate (which means "sustained initial principal in worst case").

The default setting is 60/40 (TSM / intermediate-term bonds). It shows a safe withdrawal rate of 4.5% and a perpetual withdrawal rate of 3.5% over 30 years.
I'm sorry. There must be some sort of disconnect.
...
That means average balanced portfolio returns won't be able to sustain past "safe" withdrawal rates with the same chance of success for two reasons: lower average returns and highe portfolio volatility (stock returns will be lower, but not their volatility !).
Continue to plan for 4% SWR being obvlisious of this fundamental change in financial markets and your chances of disappointment will be higher.
I agree generally with your POV. As you say we can't plausibly say that periods including the 1929-30's stock debacle and 1960's onward stagflation, US-only versions of both, are *the* worst case, pretty obviously not compared to what has happened other places. But then if we retreat to 'OK but they are the 95%-tile (or whatever) worst case' the latter statement *does* depend on the whole distribution of returns, to say the other 95% are better. And so firecalc's fatal flaw of implicitly assuming the distribution of future returns will look like the distribution of past realized returns. Which for bonds is obviously too optimistic. And also fairly clearly too optimistic for stocks IMO.

The only thing I'd quibble with is whether we actually know anything useful about the variance of the future distribution of returns. I think one could make a plausible argument that it's also lower than past, that one reason risk premia have tended to compress, along with 'riskless' rate being lower, is less risk. Though I personally wouldn't rely on that being the case.

If the distribution of returns did not itself vary over time firecalc would be a great tool, but since the distribution is almost surely changing over time it's a pretty useless tool. Simply guessing the lowest (geometric) average real return likely to prevail, thus focusing on the fact you are guessing, and *must* guess, IMO is a better way of thinking of the problem.

I'd go along with a recent post just saying 'don't try to make anything certain beyond maybe 80%', but a forward looking seat-of-pants 80%, not firecalc's 80% based on the obviously flawed implicit assumption of a stationary distribution of returns over time.

Again bottom line I don't disagree with cryptic posts like '3% is conservative, 2.2% past conservative' (though not 'irrational' necessarily, because usually it accompanies an implicit goal of inter-generational wealth transfer). I think you should be conservative about something like this.

CnC
Posts: 840
Joined: Thu May 11, 2017 12:41 pm

Re: Will we be OK at 3% WR?

Post by CnC » Fri Nov 08, 2019 4:56 pm

Thesaints wrote:
Fri Nov 08, 2019 3:46 pm
CFM300 wrote:
Thu Nov 07, 2019 9:50 pm
You might find this link interesting and useful:

https://portfoliocharts.com/portfolio/withdrawal-rates/

Not only does it explain (albeit briefly) the back-testing methodologies employed in previous studies, it has a fantastic calculator that allows you to see both the safe withdrawal rate (which means "did not run out of money in worst case") and the perpetual withdrawal rate (which means "sustained initial principal in worst case").

The default setting is 60/40 (TSM / intermediate-term bonds). It shows a safe withdrawal rate of 4.5% and a perpetual withdrawal rate of 3.5% over 30 years.
I'm sorry. There must be some sort of disconnect.

By "did not run out in the worst case" they mean the worst case "based on real-life sequence of returns". "Perpetual withdrawal rate", changes the constraint from "still having money after 30 years" (which can mean having as little as $1) to "still having the initial capital after 30 years" (which, effectively means "in perpetuity" since one can start another 30 years). This also is based on "based on real-life sequence of returns".

The Trinity Study, which many like to cite here, had a yet different (and looser) constraint "still having money after 30 years in at least x% of cases". Obviously, the Trinity Study cannot be based on the "worst possible case" historically based or not. It is instead based on all historical returns series, good and bad.

Once we accept the above, my objection is that past history average stock returns has been 10% and bond return has been 7%. This is obviously true for the 90 years of recorded history, for 40-year long periods, and for most 30-year long ones, which are relevant for retirement planning purposes. It is not true for 10-years long periods; there are many with quite different average returns, but those are not too relevant for retirement purposes.

Now, drum-roll, things are different for developed markets. They have been different for Japan for the last 30 years. They will be different for Europe and the US, until monetary policies (i.e. loose money) change and there is no reason why they should change. We can discuss this separately.
Present and foreseable monetary policies tell us that we can only dream of 10% average returns for stocks and, above all, bond returns will be close to zero real.
That means average balanced portfolio returns won't be able to sustain past "safe" withdrawal rates with the same chance of success for two reasons: lower average returns and highe portfolio volatility (stock returns will be lower, but not their volatility !).
Continue to plan for 4% SWR being oblivious of this fundamental change in financial markets and your chances of disappointment will be higher.

I really don't think you understand what others are saying. It appears that you continue to throw out 10% stocks and 7% bonds yet noone else does.

You are claiming with out proof that somehow "Present and foreseable monetary policies" will somehow result in worse returns than 2 world wars and a nuclear cold war.

I would suggest going back and reading a bit more about safe withdrawal rates. Because under the "typical" returns the portfolio is more than double what it started at.

bltn
Posts: 561
Joined: Mon Feb 20, 2017 9:32 pm

Re: Will we be OK at 3% WR?

Post by bltn » Fri Nov 08, 2019 5:20 pm

I ll try to balance several factors in my retirement spending.

1. My withdrawal rate will have to be relatively secure over my optimistic planned retirement duration of 35-40 years.
2. My retirement rate will have to allow me to enjoy some of the advantages of my accumulation in addition to security. In other words, the withdrawal rate should be high enough to allow me to spend enough to enjoy what my years off frugality have provided.
3. A legacy for my children and their children is also important to me.

To balance these three goals, I feel that a 3% annual withdrawal rate will be safe and allow a certain amount of discretionary spending.

If I had more, a lower SWR would be possible.

Garfieldthecat
Posts: 48
Joined: Mon Dec 24, 2018 10:48 am

Re: Will we be OK at 3% WR?

Post by Garfieldthecat » Fri Nov 08, 2019 5:51 pm

Rob1 wrote:
Fri Nov 08, 2019 3:33 pm
Garfieldthecat wrote:
Wed Nov 06, 2019 7:56 am
I can't find the webpage, but I've seen it linked in the forum a couple of times, where it basically says a success rate of 95% is fine, because you can't predict an asteroid crashing into us, or WWIII starting, or some other huge disaster.
"But history teaches us that depriving ourselves to boost our 40-year success probability much beyond 80% is a fool’s errand, since all you are doing is increasing the probability of failure for political, economic, and military reasons relative to the failure of banal financial planning."

More context in Bernstein‘s article here:
The Retirement Calculator from Hell, Part III
http://www.efficientfrontier.com/ef/901/hell3.htm
That's the one. Tried to find it quickly but couldn't.

User avatar
HomerJ
Posts: 13342
Joined: Fri Jun 06, 2008 12:50 pm

Re: Will we be OK at 3% WR?

Post by HomerJ » Fri Nov 08, 2019 6:22 pm

Thesaints wrote:
Fri Nov 08, 2019 3:46 pm
my objection is that past history average stock returns has been 10% and bond return has been 7%. This is obviously true for the 90 years of recorded history, for 40-year long periods, and for most 30-year long ones, which are relevant for retirement planning purposes.
Yes, for most 30-year periods it was true... And during those periods, one could pull 6%-8% a year and not run out of money.

During SOME 30-year periods, getting 10% stock returns and 7% bond returns was NOT true.. And during those bad low-return times, one could only pull 4% or 5% a year and not run out of money.

The 4% "rule" is not based on 30-year periods where we get 10% stock returns and 7% bond returns. Those conditions are not necessary for 4% withdrawals to be successful.

So stating we are not likely get 10% stock returns and 7% bond returns going forward is meaningless.

We pick 4% withdrawal rates PRECISELY because we know those kind of returns are never guaranteed. Instead we pick the worst case (so far!) historical cases, and we see that 4% worked. So if we do get low returns, like those worst-case (so far!) 30-year periods in the past, we will still be financially successful.

Now, one can make a case that the next 30 years might be worse than the U.S. historical past.. That's absolutely possible.

But you need to stop thinking that we need 10% stock returns and 7% bond returns for 4% withdrawals to work.
The J stands for Jay

jginseattle
Posts: 761
Joined: Fri Jul 01, 2011 7:33 pm

Re: Will we be OK at 3% WR?

Post by jginseattle » Fri Nov 08, 2019 6:25 pm

If your portfoliio includes risky assets will you rebalance during bad times? Investors may not match market returns.

And in this low interest rate environment I believe it's prudent to stick with a lower withdrawal rate than historical data may support.
Last edited by jginseattle on Fri Nov 08, 2019 6:27 pm, edited 1 time in total.

Thesaints
Posts: 2905
Joined: Tue Jun 20, 2017 12:25 am

Re: Will we be OK at 3% WR?

Post by Thesaints » Fri Nov 08, 2019 6:27 pm

CnC wrote:
Fri Nov 08, 2019 4:56 pm
I really don't think you understand what others are saying. It appears that you continue to throw out 10% stocks and 7% bonds yet noone else does.
When you use a 30-year period historical return to say "Don't you see ? My portfolio was not depleted" you are using an average return of about 10% for stocks and 7% for bonds. Maybe you don't know it, but that's what you are doing. There have been no 30-year long period with widely different returns, as the after-inflation table I posted shows.
You are claiming with out proof that somehow "Present and foreseable monetary policies" will somehow result in worse returns than 2 world wars and a nuclear cold war.
The actual wars lasted ~5 years each and we care about 30-year long periods. The nuclear war didn't happen at all and in fact it was a driver of market returns.
What counts is the nature of financial markets, which has changed for developed countries: population not growing any more, stock market already representing a large chunk of GDP, with the part of the economy still outside financial markets largely composed of activities that will never be part of them.
I would suggest going back and reading a bit more about safe withdrawal rates. Because under the "typical" returns the portfolio is more than double what it started at.
Unfortunately, average and typical results don't count that much in retirement planning. Typical result of a car drive is getting safely to your destination, but you still have insurance, airbags and wear a seatbelt, right ?
Last edited by Thesaints on Fri Nov 08, 2019 6:32 pm, edited 1 time in total.

Thesaints
Posts: 2905
Joined: Tue Jun 20, 2017 12:25 am

Re: Will we be OK at 3% WR?

Post by Thesaints » Fri Nov 08, 2019 6:32 pm

HomerJ wrote:
Fri Nov 08, 2019 6:22 pm
We pick 4% withdrawal rates PRECISELY because we know those kind of returns are never guaranteed. Instead we pick the worst case (so far!) historical cases, and we see that 4% worked. So if we do get low returns, like those worst-case (so far!) 30-year periods in the past, we will still be financially successful.

Now, one can make a case that the next 30 years might be worse than the U.S. historical past.. That's absolutely possible.

But you need to stop thinking that we need 10% stock returns and 7% bond returns for 4% withdrawals to work.
The study shows that there was a non-zero occurrence of failures by withdrawing 4% from a 60/40 portfolio over 30 years in the past.
During all that time average market returns (i.e. long-term averages) were higher than what we can expect for the future.
If future long-term average returns are lower, the chance of early depletion will necessarily be higher than what occurred in the past.

I'm certainly not saying that one will necessarily run out of money, just that he will be more likely to do so.

emoore
Posts: 557
Joined: Mon Mar 04, 2013 8:16 pm

Re: Will we be OK at 3% WR?

Post by emoore » Fri Nov 08, 2019 6:35 pm

Thesaints wrote:
Fri Nov 08, 2019 6:32 pm
HomerJ wrote:
Fri Nov 08, 2019 6:22 pm
We pick 4% withdrawal rates PRECISELY because we know those kind of returns are never guaranteed. Instead we pick the worst case (so far!) historical cases, and we see that 4% worked. So if we do get low returns, like those worst-case (so far!) 30-year periods in the past, we will still be financially successful.

Now, one can make a case that the next 30 years might be worse than the U.S. historical past.. That's absolutely possible.

But you need to stop thinking that we need 10% stock returns and 7% bond returns for 4% withdrawals to work.
The study shows that there was a non-zero occurrence of failures by withdrawing 4% from a 60/40 portfolio over 30 years in the past.
During all that time average market returns (i.e. long-term averages) were higher than what we can expect for the future.
If future long-term average returns are lower, the chance of early depletion will necessarily be higher than what occurred in the past.

I'm certainly not saying that one will necessarily run out of money, just that he will be more likely to do so.
There is no evidence that this will happen. It's just your opinion just like any other financial predictions of the future. I'll stick with the best real data we have and that's what happened in the past. It's the best data there is.

Thesaints
Posts: 2905
Joined: Tue Jun 20, 2017 12:25 am

Re: Will we be OK at 3% WR?

Post by Thesaints » Fri Nov 08, 2019 6:39 pm

emoore wrote:
Fri Nov 08, 2019 6:35 pm
There is no evidence that this will happen.
It is simple math. Well, maybe not "simple" as in "everyone understands it easily", but more like "everyone with a solid grasp of statistics will have no problem agreeing".

Past data is only useful if one understands the conditions that generated it. Otherwise one could simply look at the history of the Chinese stock market returns and conclude that it will always be zero, as indeed it was for more than half a century.

User avatar
willthrill81
Posts: 13594
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Will we be OK at 3% WR?

Post by willthrill81 » Fri Nov 08, 2019 6:48 pm

Thesaints wrote:
Fri Nov 08, 2019 6:32 pm
HomerJ wrote:
Fri Nov 08, 2019 6:22 pm
We pick 4% withdrawal rates PRECISELY because we know those kind of returns are never guaranteed. Instead we pick the worst case (so far!) historical cases, and we see that 4% worked. So if we do get low returns, like those worst-case (so far!) 30-year periods in the past, we will still be financially successful.

Now, one can make a case that the next 30 years might be worse than the U.S. historical past.. That's absolutely possible.

But you need to stop thinking that we need 10% stock returns and 7% bond returns for 4% withdrawals to work.
The study shows that there was a non-zero occurrence of failures by withdrawing 4% from a 60/40 portfolio over 30 years in the past.
During all that time average market returns (i.e. long-term averages) were higher than what we can expect for the future.
If future long-term average returns are lower, the chance of early depletion will necessarily be higher than what occurred in the past.

I'm certainly not saying that one will necessarily run out of money, just that he will be more likely to do so.
Average market returns mean absolutely nothing for the purpose of calculating historic safe withdrawal rates. Nothing. For instance, returns from 2017 do not enter into the calculation of the SWR for the 1966-1995 period. So you need to purge that idea from your thinking.

As others have noted, a 1.22% constant real return is all that is needed to make the '4% rule' work.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

Thesaints
Posts: 2905
Joined: Tue Jun 20, 2017 12:25 am

Re: Will we be OK at 3% WR?

Post by Thesaints » Fri Nov 08, 2019 7:34 pm

willthrill81 wrote:
Fri Nov 08, 2019 6:48 pm
Average market returns mean absolutely nothing for the purpose of calculating historic safe withdrawal rates. Nothing. For instance, returns from 2017 do not enter into the calculation of the SWR for the 1966-1995 period. So you need to purge that idea from your thinking.
Maybe you skipped where I wrote that 30-years average returns have never been too different from 10/7.5 (especially after factoring in inflation). There is a table I linked in the previous page showing that 1966-1995 was typical.
As others have noted, a 1.22% constant real return is all that is needed to make the '4% rule' work.
Yep. "constant" and "work" are they keywords. "Constant" you won't get and "work" may mean than on year 29 you have 4% of your inflation-adjusted initial capital, plus $1. Would that work for you ?

User avatar
FiveK
Posts: 7741
Joined: Sun Mar 16, 2014 2:43 pm

Re: Will we be OK at 3% WR?

Post by FiveK » Fri Nov 08, 2019 7:54 pm

Thesaints wrote:
Fri Nov 08, 2019 7:34 pm
Maybe you skipped where I wrote that 30-years average returns have never been too different from 10/7.5 (especially after factoring in inflation).
Seems you are saying "this time it's different." Is that correct?
Thesaints wrote:
Fri Nov 08, 2019 4:04 pm
The lowest (of all!) 30-year period is 4.3%. Japanese last 30-year return ? Less than zero.
Are you suggesting that one should use a negative return on investments for planning purposes?

What do you consider a Safe Withdrawal Rate (one having, say, a 90% chance of being "safe")?

User avatar
willthrill81
Posts: 13594
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Will we be OK at 3% WR?

Post by willthrill81 » Fri Nov 08, 2019 8:04 pm

Thesaints wrote:
Fri Nov 08, 2019 7:34 pm
willthrill81 wrote:
Fri Nov 08, 2019 6:48 pm
Average market returns mean absolutely nothing for the purpose of calculating historic safe withdrawal rates. Nothing. For instance, returns from 2017 do not enter into the calculation of the SWR for the 1966-1995 period. So you need to purge that idea from your thinking.
Maybe you skipped where I wrote that 30-years average returns have never been too different from 10/7.5 (especially after factoring in inflation). There is a table I linked in the previous page showing that 1966-1995 was typical.
That's false on two levels. First, 30 year returns have not been predictive at all of the SWR of that 30 year period. The graph below from poster Siamond displays this very well. Second, the graph below also demonstrates that portfolio returns over historic 30 year periods were often well below the 7-10% range you refer to. In fact, I believe that there were only about 3 historic periods where the average real return on a 60/40 portfolio was just about 7%. In most instances, the average real return was between 4-6.5%. Yet observe that several periods had average real returns of about 4% or even lower but had a SWR of 7% or higher.
siamond wrote:
Wed Feb 06, 2019 3:35 pm
Note that the relation between CAGR and SWR is surprisingly loose, as Will already hinted at. Let me try to make the point more clear because it is certainly a tad counter-intuitive. I assembled the following graphs by tweaking a tad the Simba backtesting spreadsheet. Every vertical is a retirement cycle of 30 years, starting on the year indicated on the X axis. And you'll find the max WR (aka SWR) for each individual cycle as well as the (real) CAGR. Check those graphs carefully. If you perceive that the blue points (CAGR) give you any solid indication about the lowest pink points (SWR), I don't know, look again...

Image

Image
This is because a poor sequence of returns in the beginning of the period brings down the SWR for the entire period. As I have said at least once in this thread already, Kitces found back in 2012 that 30 year SWRs were highly correlated (r = .91) with the returns of the first 15 years of the 30 year period. And the first 15 years of returns associated with the lowest historic SWRs (i.e. 4%) were less than 1% real for 60/40 portfolios.

And again, none of this should matter much because it's entirely unnecessary for virtually anyone to actually employ a SWR approach to make withdrawals. There are much better methods out there.
Last edited by willthrill81 on Fri Nov 08, 2019 8:13 pm, edited 2 times in total.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

EddyB
Posts: 1039
Joined: Fri May 24, 2013 3:43 pm

Re: Will we be OK at 3% WR?

Post by EddyB » Fri Nov 08, 2019 8:12 pm

willthrill81 wrote:
Fri Nov 08, 2019 8:04 pm
Thesaints wrote:
Fri Nov 08, 2019 7:34 pm
willthrill81 wrote:
Fri Nov 08, 2019 6:48 pm
Average market returns mean absolutely nothing for the purpose of calculating historic safe withdrawal rates. Nothing. For instance, returns from 2017 do not enter into the calculation of the SWR for the 1966-1995 period. So you need to purge that idea from your thinking.
Maybe you skipped where I wrote that 30-years average returns have never been too different from 10/7.5 (especially after factoring in inflation). There is a table I linked in the previous page showing that 1966-1995 was typical.
That's false on two levels. First, 30 year returns have not been predictive at all of the SWR of that 30 year period. The graph below from poster Siamond displays this very well. Second, the graph below also demonstrates that portfolio returns over historic 30 year periods were often well below the 7-10% range you refer to. In fact, I believe that there were only about 3 historic periods where the average real return on a 60/40 portfolio was just about 7%. In most instances, the average real return was between 4-6.5%. Yet observe how many periods had average real returns of about 4% or even lower but had a SWR of 7% or higher.
siamond wrote:
Wed Feb 06, 2019 3:35 pm
Note that the relation between CAGR and SWR is surprisingly loose, as Will already hinted at. Let me try to make the point more clear because it is certainly a tad counter-intuitive. I assembled the following graphs by tweaking a tad the Simba backtesting spreadsheet. Every vertical is a retirement cycle of 30 years, starting on the year indicated on the X axis. And you'll find the max WR (aka SWR) for each individual cycle as well as the (real) CAGR. Check those graphs carefully. If you perceive that the blue points (CAGR) give you any solid indication about the lowest pink points (SWR), I don't know, look again...

Image

Image
This is because a poor sequence of returns in the beginning of the period brings down the SWR for the entire period. As I have said at least once in this thread already, Kitces found back in 2012 that 30 year SWRs were highly correlated (r = .91) with the returns of the first 15 years of the 30 year period. And the first 15 years of returns associated with the lowest historic SWRs (i.e. 4%) were less than 1% real for 60/40 portfolios.
I commend your patience.

Thesaints
Posts: 2905
Joined: Tue Jun 20, 2017 12:25 am

Re: Will we be OK at 3% WR?

Post by Thesaints » Fri Nov 08, 2019 8:31 pm

willthrill81 wrote:
Fri Nov 08, 2019 8:04 pm
That's false on two levels. First, 30 year returns have not been predictive at all of the SWR of that 30 year period.
We all know that: volatility works against you during withdrawal. So, if you have a constant x% return, you can certainly safely withdraw x% and even more, if you accept to deplete your capital. Once you add volatility, your allowed withdrawal rate goes down.
In fact, I believe that there were only about 3 historic periods where the average real return on a 60/40 portfolio was just about 7%. In most instances, the average real return was between 4-6.5%. Yet observe that several periods had average real returns of about 4% or even lower but had a SWR of 7% or higher.
We might be confusing each other by mixing real and nominal returns. However, the 90+ year historical return calculated on the 90+ years we have more or less homogeneous data available show that average nominal returns stock/bonds have been 10/7.5 (I might be off a few decimals). With that in mind there is no way in heck for almost all 30-year periods to have had lower returns. It is simple math. The table I linked shows that the ex-inflation return of all 30-year periods has been fairly constant, within a couple % spread (the blue points on your plot).

Note that the relation between CAGR and SWR is surprisingly loose, as Will already hinted at. Let me try to make the point more clear because it is certainly a tad counter-intuitive. I assembled the following graphs by tweaking a tad the Simba backtesting spreadsheet. Every vertical is a retirement cycle of 30 years, starting on the year indicated on the X axis. And you'll find the max WR (aka SWR) for each individual cycle as well as the (real) CAGR. Check those graphs carefully. If you perceive that the blue points (CAGR) give you any solid indication about the lowest pink points (SWR), I don't know, look again...
Not surprising at all. It is simply the effect of volatility within each 30-year period, which is not shown on your plot.
And again, none of this should matter much because it's entirely unnecessary for virtually anyone to actually employ a SWR approach to make withdrawals. There are much better methods out there.
Perhaps, but the original problem is "What is the maximum constant withdrawal I can make for N years, given an initial portfolio ?". That question is a rather natural and relevant one.

marcopolo
Posts: 2406
Joined: Sat Dec 03, 2016 10:22 am

Re: Will we be OK at 3% WR?

Post by marcopolo » Fri Nov 08, 2019 9:04 pm

EddyB wrote:
Fri Nov 08, 2019 8:12 pm
willthrill81 wrote:
Fri Nov 08, 2019 8:04 pm
Thesaints wrote:
Fri Nov 08, 2019 7:34 pm
willthrill81 wrote:
Fri Nov 08, 2019 6:48 pm
Average market returns mean absolutely nothing for the purpose of calculating historic safe withdrawal rates. Nothing. For instance, returns from 2017 do not enter into the calculation of the SWR for the 1966-1995 period. So you need to purge that idea from your thinking.
Maybe you skipped where I wrote that 30-years average returns have never been too different from 10/7.5 (especially after factoring in inflation). There is a table I linked in the previous page showing that 1966-1995 was typical.
That's false on two levels. First, 30 year returns have not been predictive at all of the SWR of that 30 year period. The graph below from poster Siamond displays this very well. Second, the graph below also demonstrates that portfolio returns over historic 30 year periods were often well below the 7-10% range you refer to. In fact, I believe that there were only about 3 historic periods where the average real return on a 60/40 portfolio was just about 7%. In most instances, the average real return was between 4-6.5%. Yet observe how many periods had average real returns of about 4% or even lower but had a SWR of 7% or higher.
siamond wrote:
Wed Feb 06, 2019 3:35 pm
Note that the relation between CAGR and SWR is surprisingly loose, as Will already hinted at. Let me try to make the point more clear because it is certainly a tad counter-intuitive. I assembled the following graphs by tweaking a tad the Simba backtesting spreadsheet. Every vertical is a retirement cycle of 30 years, starting on the year indicated on the X axis. And you'll find the max WR (aka SWR) for each individual cycle as well as the (real) CAGR. Check those graphs carefully. If you perceive that the blue points (CAGR) give you any solid indication about the lowest pink points (SWR), I don't know, look again...

Image

Image
This is because a poor sequence of returns in the beginning of the period brings down the SWR for the entire period. As I have said at least once in this thread already, Kitces found back in 2012 that 30 year SWRs were highly correlated (r = .91) with the returns of the first 15 years of the 30 year period. And the first 15 years of returns associated with the lowest historic SWRs (i.e. 4%) were less than 1% real for 60/40 portfolios.
I commend your patience.
Agreed. Several people are providing a great service is explaining this. I admittedly gave up earlier in the thread.
Once in a while you get shown the light, in the strangest of places if you look at it right.

D-Dog
Posts: 73
Joined: Sun Dec 02, 2007 12:44 am

Re: Will we be OK at 3% WR?

Post by D-Dog » Fri Nov 08, 2019 11:42 pm

marcopolo wrote:
Fri Nov 08, 2019 9:04 pm
EddyB wrote:
Fri Nov 08, 2019 8:12 pm
willthrill81 wrote:
Fri Nov 08, 2019 8:04 pm
Thesaints wrote:
Fri Nov 08, 2019 7:34 pm
willthrill81 wrote:
Fri Nov 08, 2019 6:48 pm
Average market returns mean absolutely nothing for the purpose of calculating historic safe withdrawal rates. Nothing. For instance, returns from 2017 do not enter into the calculation of the SWR for the 1966-1995 period. So you need to purge that idea from your thinking.
Maybe you skipped where I wrote that 30-years average returns have never been too different from 10/7.5 (especially after factoring in inflation). There is a table I linked in the previous page showing that 1966-1995 was typical.
That's false on two levels. First, 30 year returns have not been predictive at all of the SWR of that 30 year period. The graph below from poster Siamond displays this very well. Second, the graph below also demonstrates that portfolio returns over historic 30 year periods were often well below the 7-10% range you refer to. In fact, I believe that there were only about 3 historic periods where the average real return on a 60/40 portfolio was just about 7%. In most instances, the average real return was between 4-6.5%. Yet observe how many periods had average real returns of about 4% or even lower but had a SWR of 7% or higher.
siamond wrote:
Wed Feb 06, 2019 3:35 pm
Note that the relation between CAGR and SWR is surprisingly loose, as Will already hinted at. Let me try to make the point more clear because it is certainly a tad counter-intuitive. I assembled the following graphs by tweaking a tad the Simba backtesting spreadsheet. Every vertical is a retirement cycle of 30 years, starting on the year indicated on the X axis. And you'll find the max WR (aka SWR) for each individual cycle as well as the (real) CAGR. Check those graphs carefully. If you perceive that the blue points (CAGR) give you any solid indication about the lowest pink points (SWR), I don't know, look again...

Image

Image
This is because a poor sequence of returns in the beginning of the period brings down the SWR for the entire period. As I have said at least once in this thread already, Kitces found back in 2012 that 30 year SWRs were highly correlated (r = .91) with the returns of the first 15 years of the 30 year period. And the first 15 years of returns associated with the lowest historic SWRs (i.e. 4%) were less than 1% real for 60/40 portfolios.
I commend your patience.
Agreed. Several people are providing a great service is explaining this. I admittedly gave up earlier in the thread.
I think Thesaints might be trying to say this: the 4% SWR is based on a time period with a certain distribution of returns. If (and it’s a big if IMO) the future long term distribution of returns is lower (e.g., the whole distribution is shifted left), then the future worst case scenario gets worse, the future best case scenario gets worse, and the future average scenario gets worse. And, in particular, a 4% withdrawal rate, which may have a 99% probability of success in a world where the historical return distribution applies may have a lower probability of success in a world where a different return distribution applies. I think this is a logical conclusion if you agree with the premise that the future distribution of returns will be lower. If this is what Thesaints is trying to say, then I think the discussion should really be about debating the premise of whether the future long term distribution of returns will be different. I’m personally not convinced that it will be, but I’m just trying to help make sense of the discussion so that it can be more productive.

Topic Author
bhsince87
Posts: 2551
Joined: Thu Oct 03, 2013 1:08 pm

Re: Will we be OK at 3% WR?

Post by bhsince87 » Sat Nov 09, 2019 12:19 am

OP here, and one more thing I will add is portfolio mix. That is definitely a relevant factor I left out.

I retired at 55/45, with a plan of letting it ride up to 60/40.

After ten months, it's already at 56/44!

I figure spending bonds early on is the best way to combat sequence of return risks.

But no mention of Social Security or other pensions, or total net worth. I don't see how they are relevant in a SWR discussion.
"If ye love wealth better than liberty, the tranquility of servitude better than the animating contest of freedom, go home from us in peace." Samuel Adams

konic
Posts: 23
Joined: Fri Jan 05, 2018 5:26 pm

Re: Will we be OK at 3% WR?

Post by konic » Sat Nov 09, 2019 1:14 am

D-Dog wrote:
Fri Nov 08, 2019 11:42 pm
marcopolo wrote:
Fri Nov 08, 2019 9:04 pm
EddyB wrote:
Fri Nov 08, 2019 8:12 pm
willthrill81 wrote:
Fri Nov 08, 2019 8:04 pm
Thesaints wrote:
Fri Nov 08, 2019 7:34 pm

Maybe you skipped where I wrote that 30-years average returns have never been too different from 10/7.5 (especially after factoring in inflation). There is a table I linked in the previous page showing that 1966-1995 was typical.
That's false on two levels. First, 30 year returns have not been predictive at all of the SWR of that 30 year period. The graph below from poster Siamond displays this very well. Second, the graph below also demonstrates that portfolio returns over historic 30 year periods were often well below the 7-10% range you refer to. In fact, I believe that there were only about 3 historic periods where the average real return on a 60/40 portfolio was just about 7%. In most instances, the average real return was between 4-6.5%. Yet observe how many periods had average real returns of about 4% or even lower but had a SWR of 7% or higher.
siamond wrote:
Wed Feb 06, 2019 3:35 pm
Note that the relation between CAGR and SWR is surprisingly loose, as Will already hinted at. Let me try to make the point more clear because it is certainly a tad counter-intuitive. I assembled the following graphs by tweaking a tad the Simba backtesting spreadsheet. Every vertical is a retirement cycle of 30 years, starting on the year indicated on the X axis. And you'll find the max WR (aka SWR) for each individual cycle as well as the (real) CAGR. Check those graphs carefully. If you perceive that the blue points (CAGR) give you any solid indication about the lowest pink points (SWR), I don't know, look again...

Image

Image
This is because a poor sequence of returns in the beginning of the period brings down the SWR for the entire period. As I have said at least once in this thread already, Kitces found back in 2012 that 30 year SWRs were highly correlated (r = .91) with the returns of the first 15 years of the 30 year period. And the first 15 years of returns associated with the lowest historic SWRs (i.e. 4%) were less than 1% real for 60/40 portfolios.
I commend your patience.
Agreed. Several people are providing a great service is explaining this. I admittedly gave up earlier in the thread.
I think Thesaints might be trying to say this: the 4% SWR is based on a time period with a certain distribution of returns. If (and it’s a big if IMO) the future long term distribution of returns is lower (e.g., the whole distribution is shifted left), then the future worst case scenario gets worse, the future best case scenario gets worse, and the future average scenario gets worse. And, in particular, a 4% withdrawal rate, which may have a 99% probability of success in a world where the historical return distribution applies may have a lower probability of success in a world where a different return distribution applies. I think this is a logical conclusion if you agree with the premise that the future distribution of returns will be lower. If this is what Thesaints is trying to say, then I think the discussion should really be about debating the premise of whether the future long term distribution of returns will be different. I’m personally not convinced that it will be, but I’m just trying to help make sense of the discussion so that it can be more productive.
I think you are giving too much credit to TheSaints. The person seems to intentionally want to be dense in understand a simple argument about the basis for the 4% SWR. I do not think there is much point in trying to explain anything to this person.

Moderators... this thread has run its course, even the OP has already chimed-in essentially indicating they got sufficient inputs to proceed with their course.

What a pointless exercise this is turning into :oops:

User avatar
HomerJ
Posts: 13342
Joined: Fri Jun 06, 2008 12:50 pm

Re: Will we be OK at 3% WR?

Post by HomerJ » Sat Nov 09, 2019 1:29 am

EddyB wrote:
Fri Nov 08, 2019 8:12 pm
I commend your patience.
I'm done.
The J stands for Jay

User avatar
willthrill81
Posts: 13594
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Will we be OK at 3% WR?

Post by willthrill81 » Sat Nov 09, 2019 9:37 am

Thesaints wrote:
Fri Nov 08, 2019 8:31 pm
Perhaps, but the original problem is "What is the maximum constant withdrawal I can make for N years, given an initial portfolio ?". That question is a rather natural and relevant one.
You shifting the goalposts again, but I would argue that the question is largely irrelevant because virtually no one is actually employing a SWR approach, nor should they. Doing so would introduce a non-zero risk of portfolio depletion and potentially a much greater risk of not using available resources (i.e. richest person the cemetery). It may be the worst withdrawal approach out there, and even its 'founder', Bill Bengen, never argued to my knowledge that anyone actually use it. He merely argued that those anticipating a ~30 year retirement shouldn't be withdrawing 7% from their portfolio to begin with, more like 4%, which they could subsequently adjust based on portfolio performance. Used in such a manner, beginning one's withdrawals at 4%, assuming that it is appropriate for one's life expectancy and other goals, is plausible. But deciding on a fixed inflation-adjusted dollar amount to withdraw from one's portfolio for the next 30 years, come hell or high water, would be absolutely ridiculous and reckless.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

JackoC
Posts: 1005
Joined: Sun Aug 12, 2018 11:14 am

Re: Will we be OK at 3% WR?

Post by JackoC » Sat Nov 09, 2019 9:45 am

D-Dog wrote:
Fri Nov 08, 2019 11:42 pm

I think Thesaints might be trying to say this: the 4% SWR is based on a time period with a certain distribution of returns. If (and it’s a big if IMO) the future long term distribution of returns is lower (e.g., the whole distribution is shifted left), then the future worst case scenario gets worse, the future best case scenario gets worse, and the future average scenario gets worse. And, in particular, a 4% withdrawal rate, which may have a 99% probability of success in a world where the historical return distribution applies may have a lower probability of success in a world where a different return distribution applies. I think this is a logical conclusion if you agree with the premise that the future distribution of returns will be lower. If this is what Thesaints is trying to say, then I think the discussion should really be about debating the premise of whether the future long term distribution of returns will be different. I’m personally not convinced that it will be, but I’m just trying to help make sense of the discussion so that it can be more productive.
I think that's Thesaints point, and despite people teasing with stuff like 'I admire your patience' as if Thesaints is completely full of it...he/she is not at all IMO.

1. no reason to think the future distribution of returns from today's asset valuations is the same as unconditional past distribution of returns, a 100% obviously wrong assumption for 'riskless' bonds, and no compelling reason to believe it for any asset class. But it's the basic assumption of firecalc.

2. there isn't really that much data, just a few *independent* 30-40 yr trials in a period where the world is anything like now, even aside from Thesaints theory of now being significantly different in exogenous factors (monetary policy, low population growth, etc) from the past century or so. You don't even have to agree with that to observe how little *independent* data there is from the past, and how basically flawed it is to draw conclusions from overlapping thus autocorrelated data under the assumption it's independent.

Again the point where I'm not sure I agree with Thesaints is that we can say the variance of future return is now about the same as past variance, so the whole distribution of similar shape just moves down to account for today's higher asset valuations (lower bond and earnings yields). We don't actually know that either. But assuming there is less risk now therefore somewhat offsetting the shift down in the mean of the distribution in terms of generating worst case future paths is an obviously risky assumption.

Finally the repeated statements that 'worst cases' have nothing to do with the whole distribution are also obviously wrong, no matter how much group-think '+1-ing' there is. Nobody knows *the* worst case or even pretends to, to be fair. The idea of firecalc is that we can quantify the %-tile likelihood of a given bad case based on history (albeit using a flawed statistical method of overlapping therefore autocorrelated periods). But if you say X path is the 95%-tile worst, you're automatically depending on the rest of the distribution, to say the other 95% of cases are not as bad or worse. And in the historical data set they 'aren't as bad' because they average something like 7% real return on stocks, and also much higher than today's 0.6% expected 30 yr real return on 'riskless' bonds. IOW the reason so few are really bad is that they average being very good compared to a realistic estimate of expected return *from now*. :happy If they average a lot lower than that, more of them would be as bad as the past 95% worst case, assuming the future variance of return is same as past variance. Again we don't know that future variance is same as past variance, but firecalc tells us nothing about that either.

Again, it's simple math to say 4% SWR lasts 29 yrs at 1% constant real return. And 1% constant return is low. Even with 'sequence of return' effect...there's also 'cutting back somewhat from straight SWR if/when necessary' effect. That doesn't look so bad. But the % likelihood of success of that strategy from tools like firecalc is overstated, for the simple reason that those outputs are based on a distribution of returns probably quite a bit higher than the distribution we're looking at now given now's valuations and the whole distribution *does* affect what *future* %-tile the *past* 'worst case' appears to be. That doesn't mean 4% is a crazy strategy necessarily, it's just not as conservative as the limited statistical history makes it look.

visualguy
Posts: 1600
Joined: Thu Jan 30, 2014 1:32 am

Re: Will we be OK at 3% WR?

Post by visualguy » Sat Nov 09, 2019 10:19 am

willthrill81 wrote:
Sat Nov 09, 2019 9:37 am
Thesaints wrote:
Fri Nov 08, 2019 8:31 pm
Perhaps, but the original problem is "What is the maximum constant withdrawal I can make for N years, given an initial portfolio ?". That question is a rather natural and relevant one.
You shifting the goalposts again, but I would argue that the question is largely irrelevant because virtually no one is actually employing a SWR approach, nor should they. Doing so would introduce a non-zero risk of portfolio depletion and potentially a much greater risk of not using available resources (i.e. richest person the cemetery). It may be the worst withdrawal approach out there, and even its 'founder', Bill Bengen, never argued to my knowledge that anyone actually use it. He merely argued that those anticipating a ~30 year retirement shouldn't be withdrawing 7% from their portfolio to begin with, more like 4%, which they could subsequently adjust based on portfolio performance. Used in such a manner, beginning one's withdrawals at 4%, assuming that it is appropriate for one's life expectancy and other goals, is plausible. But deciding on a fixed inflation-adjusted dollar amount to withdraw from one's portfolio for the next 30 years, come hell or high water, would be absolutely ridiculous and reckless.
If a constant 4% inflation-adjusted over 30 years is now "absolutely ridiculous and reckless", that's very bad news because I think many of us assume that this is safe.

It's reasonable to ask what fixed WR wouldn't be reckless. Also, reasonable to ask how much lower than 4%+inflation one might need to go at times of poor portfolio performance. This isn't really different from planning for a lower SWR, and adding a bunch of highly discretionary items that can be cut. Same questions for 40 years instead of 30...

smitcat
Posts: 4193
Joined: Mon Nov 07, 2016 10:51 am

Re: Will we be OK at 3% WR?

Post by smitcat » Sat Nov 09, 2019 10:44 am

visualguy wrote:
Sat Nov 09, 2019 10:19 am
willthrill81 wrote:
Sat Nov 09, 2019 9:37 am
Thesaints wrote:
Fri Nov 08, 2019 8:31 pm
Perhaps, but the original problem is "What is the maximum constant withdrawal I can make for N years, given an initial portfolio ?". That question is a rather natural and relevant one.
You shifting the goalposts again, but I would argue that the question is largely irrelevant because virtually no one is actually employing a SWR approach, nor should they. Doing so would introduce a non-zero risk of portfolio depletion and potentially a much greater risk of not using available resources (i.e. richest person the cemetery). It may be the worst withdrawal approach out there, and even its 'founder', Bill Bengen, never argued to my knowledge that anyone actually use it. He merely argued that those anticipating a ~30 year retirement shouldn't be withdrawing 7% from their portfolio to begin with, more like 4%, which they could subsequently adjust based on portfolio performance. Used in such a manner, beginning one's withdrawals at 4%, assuming that it is appropriate for one's life expectancy and other goals, is plausible. But deciding on a fixed inflation-adjusted dollar amount to withdraw from one's portfolio for the next 30 years, come hell or high water, would be absolutely ridiculous and reckless.
If a constant 4% inflation-adjusted over 30 years is now "absolutely ridiculous and reckless", that's very bad news because I think many of us assume that this is safe.

It's reasonable to ask what fixed WR wouldn't be reckless. Also, reasonable to ask how much lower than 4%+inflation one might need to go at times of poor portfolio performance. This isn't really different from planning for a lower SWR, and adding a bunch of highly discretionary items that can be cut. Same questions for 40 years instead of 30...
"Also, reasonable to ask how much lower than 4%+inflation one might need to go at times of poor portfolio performance. This isn't really different from planning for a lower SWR, and adding a bunch of highly discretionary items that can be cut. Same questions for 40 years instead of 30..."

I believe this is a very valid question to ask.
After reading this forum for quite a while it became apparent that there are many talented, intelligent, unselfish, and informational folks on this board.
But there is something else that I see when taking all of these posts into account - most will 'make' their own version of what is considered reckless or not based upon their own means and plans.
Some factors that sway these decisions that are subjective.
- how much do they hate working
- what anyone considers as "discretionary"
- how well they have planned for future costs
- whether they have a backstop in other family or not
- whether they view a ridged or flexible SWR
- what their views are on LTC and medical costs
Should most of these affect the views of SWR ? Obviously not , but the opinions that will get expressed will be greatly affected by these factors.
On these posts we get through the factual information quite clearly and early and them these other factors come into play.
YMMV

EddyB
Posts: 1039
Joined: Fri May 24, 2013 3:43 pm

Re: Will we be OK at 3% WR?

Post by EddyB » Sat Nov 09, 2019 11:12 am

JackoC wrote:
Sat Nov 09, 2019 9:45 am
D-Dog wrote:
Fri Nov 08, 2019 11:42 pm

I think Thesaints might be trying to say this: the 4% SWR is based on a time period with a certain distribution of returns. If (and it’s a big if IMO) the future long term distribution of returns is lower (e.g., the whole distribution is shifted left), then the future worst case scenario gets worse, the future best case scenario gets worse, and the future average scenario gets worse. And, in particular, a 4% withdrawal rate, which may have a 99% probability of success in a world where the historical return distribution applies may have a lower probability of success in a world where a different return distribution applies. I think this is a logical conclusion if you agree with the premise that the future distribution of returns will be lower. If this is what Thesaints is trying to say, then I think the discussion should really be about debating the premise of whether the future long term distribution of returns will be different. I’m personally not convinced that it will be, but I’m just trying to help make sense of the discussion so that it can be more productive.
I think that's Thesaints point, and despite people teasing with stuff like 'I admire your patience' as if Thesaints is completely full of it...he/she is not at all IMO.

1. no reason to think the future distribution of returns from today's asset valuations is the same as unconditional past distribution of returns, a 100% obviously wrong assumption for 'riskless' bonds, and no compelling reason to believe it for any asset class. But it's the basic assumption of firecalc.

2. there isn't really that much data, just a few *independent* 30-40 yr trials in a period where the world is anything like now, even aside from Thesaints theory of now being significantly different in exogenous factors (monetary policy, low population growth, etc) from the past century or so. You don't even have to agree with that to observe how little *independent* data there is from the past, and how basically flawed it is to draw conclusions from overlapping thus autocorrelated data under the assumption it's independent.

Again the point where I'm not sure I agree with Thesaints is that we can say the variance of future return is now about the same as past variance, so the whole distribution of similar shape just moves down to account for today's higher asset valuations (lower bond and earnings yields). We don't actually know that either. But assuming there is less risk now therefore somewhat offsetting the shift down in the mean of the distribution in terms of generating worst case future paths is an obviously risky assumption.

Finally the repeated statements that 'worst cases' have nothing to do with the whole distribution are also obviously wrong, no matter how much group-think '+1-ing' there is. Nobody knows *the* worst case or even pretends to, to be fair. The idea of firecalc is that we can quantify the %-tile likelihood of a given bad case based on history (albeit using a flawed statistical method of overlapping therefore autocorrelated periods). But if you say X path is the 95%-tile worst, you're automatically depending on the rest of the distribution, to say the other 95% of cases are not as bad or worse. And in the historical data set they 'aren't as bad' because they average something like 7% real return on stocks, and also much higher than today's 0.6% expected 30 yr real return on 'riskless' bonds. IOW the reason so few are really bad is that they average being very good compared to a realistic estimate of expected return *from now*. :happy If they average a lot lower than that, more of them would be as bad as the past 95% worst case, assuming the future variance of return is same as past variance. Again we don't know that future variance is same as past variance, but firecalc tells us nothing about that either.

Again, it's simple math to say 4% SWR lasts 29 yrs at 1% constant real return. And 1% constant return is low. Even with 'sequence of return' effect...there's also 'cutting back somewhat from straight SWR if/when necessary' effect. That doesn't look so bad. But the % likelihood of success of that strategy from tools like firecalc is overstated, for the simple reason that those outputs are based on a distribution of returns probably quite a bit higher than the distribution we're looking at now given now's valuations and the whole distribution *does* affect what *future* %-tile the *past* 'worst case' appears to be. That doesn't mean 4% is a crazy strategy necessarily, it's just not as conservative as the limited statistical history makes it look.
In my opinion, what all that comes down to is the worry that the retirement period that any of us will face (the only one that matters) may be worse than any in the data set. Of course that''s possible, but I see no convincing current-valuations-based evidence presented for that view over such long time periods (especially when one considers that much of the audience is not in retirement now or expecting to commence it in the near term), and so it just becomes a matter of faith or pessimism. Even if some pessimism is reasonable, the magnitude some people suggest (that withdrawal rates may be little more than half the worst rate in the data set) is staggering and comes at truly enormous certain costs. If these proponents want to just say that simply---"I think the situation over any period of 30-50 years in the next 30-70 years could be radically worse than any time since the dawn on the twentieth century," and not claim and numeric accuracy to it, well, then people can see that as what it is, but in my view that's all it is.

tibbitts
Posts: 9317
Joined: Tue Feb 27, 2007 6:50 pm

Re: Will we be OK at 3% WR?

Post by tibbitts » Sat Nov 09, 2019 11:29 am

CnC wrote:
Fri Nov 08, 2019 2:02 pm
Thesaints wrote:
Thu Nov 07, 2019 8:19 pm
"If history is any guide for the future, then withdrawal rates of 3% and 4% are extremely unlikely to exhaust any portfolio of stocks and bonds during any of the payout periods shown in Table 1. In those cases, portfolio success seems close to being assured."

Throughout the period used for backtesting stocks return was 10% and bond returns 7%: History is not a good guide for the future anymore.
You are absolutely right, I mean everyone knew in 2009 there was no way they would be getting 10% stock returns over the next 10 years...
That's my point: that Bogleheads haven't exactly been jumping on board with dramatically increasing the recommended SWR after market drops. If you have decided on 4%, and had $1M in 2007 from which you didn't take any withdrawals, but the markets dropped your investments to $700k two years later, nobody was telling you that in 2009 you could start your retirement and take $40k/year plus inflation adjustments (so you'd actually be starting with more than $40k) from your $700k.

User avatar
willthrill81
Posts: 13594
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Will we be OK at 3% WR?

Post by willthrill81 » Sat Nov 09, 2019 11:40 am

visualguy wrote:
Sat Nov 09, 2019 10:19 am
willthrill81 wrote:
Sat Nov 09, 2019 9:37 am
Thesaints wrote:
Fri Nov 08, 2019 8:31 pm
Perhaps, but the original problem is "What is the maximum constant withdrawal I can make for N years, given an initial portfolio ?". That question is a rather natural and relevant one.
You shifting the goalposts again, but I would argue that the question is largely irrelevant because virtually no one is actually employing a SWR approach, nor should they. Doing so would introduce a non-zero risk of portfolio depletion and potentially a much greater risk of not using available resources (i.e. richest person the cemetery). It may be the worst withdrawal approach out there, and even its 'founder', Bill Bengen, never argued to my knowledge that anyone actually use it. He merely argued that those anticipating a ~30 year retirement shouldn't be withdrawing 7% from their portfolio to begin with, more like 4%, which they could subsequently adjust based on portfolio performance. Used in such a manner, beginning one's withdrawals at 4%, assuming that it is appropriate for one's life expectancy and other goals, is plausible. But deciding on a fixed inflation-adjusted dollar amount to withdraw from one's portfolio for the next 30 years, come hell or high water, would be absolutely ridiculous and reckless.
If a constant 4% inflation-adjusted over 30 years is now "absolutely ridiculous and reckless", that's very bad news because I think many of us assume that this is safe.
The issue is that a fixed real dollar withdrawal approach is inherently ridiculous and reckless. As I noted above, it needlessly introduces a non-zero chance of portfolio depletion and a far more likely chance of dying with a huge pile of unspent money. Bengen himself never suggested that anyone strictly adhere to such an approach and advocated some degree of flexibility instead. What he suggested that is 4% of one's starting portfolio balance is a good starting point for one's withdrawals, and subsequent withdrawals can be adjusted on an as-warranted basis.
visualguy wrote:
Sat Nov 09, 2019 10:19 am
Also, reasonable to ask how much lower than 4%+inflation one might need to go at times of poor portfolio performance. This isn't really different from planning for a lower SWR, and adding a bunch of highly discretionary items that can be cut. Same questions for 40 years instead of 30...
Deciding to start with a lower than 4% withdrawal rate is qualitatively distinct from preparing for the historically tiny possibility of needing to reduce one's withdrawals even further. For instance, we plan on at least half of our retirement spending being discretionary. Reducing that 10-15 years into retirement if necessary is very different from reducing that right from the start and/or delaying retirement in an effort to reduce the historically very small possibility (though never eliminating it entirely since that is impossible) of needing to reduce my discretionary spending later? Such a risk/reward trade-off seems to be very poor.
EddyB wrote:
Sat Nov 09, 2019 11:12 am
In my opinion, what all that comes down to is the worry that the retirement period that any of us will face (the only one that matters) may be worse than any in the data set. Of course that''s possible, but I see no convincing current-valuations-based evidence presented for that view over such long time periods (especially when one considers that much of the audience is not in retirement now or expecting to commence it in the near term), and so it just becomes a matter of faith or pessimism. Even if some pessimism is reasonable, the magnitude some people suggest (that withdrawal rates may be little more than half the worst rate in the data set) is staggering and comes at truly enormous certain costs. If these proponents want to just say that simply---"I think the situation over any period of 30-50 years in the next 30-70 years could be radically worse than any time since the dawn on the twentieth century," and not claim and numeric accuracy to it, well, then people can see that as what it is, but in my view that's all it is.
:thumbsup

Why would I take preemptively medicine for an illness that is historically very unlikely to occur when the medicine is expensive in terms of time lost and/or possibly reduces my quality of life, especially if contracting the illness (i.e. reducing one's discretionary spending) is far from fatal?
Last edited by willthrill81 on Sat Nov 09, 2019 11:51 am, edited 1 time in total.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

tvubpwcisla
Posts: 9
Joined: Sat Nov 09, 2019 10:09 am

Re: Will we be OK at 3% WR?

Post by tvubpwcisla » Sat Nov 09, 2019 11:51 am

Keeping your expenses as low as possible will help you fluctuate your withdrawal rate between a range that you feel comfortable with. Where you might run into some trouble is if your expenses dictate what your withdrawal rate needs to be. During a market correction, it is best to be in a place where you have a solid amount of cash on hand matched together with low monthly expense obligations. The folks that did extremely well during previous downturns in the economy and financial markets where the ones in exactly that place.

User avatar
willthrill81
Posts: 13594
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Will we be OK at 3% WR?

Post by willthrill81 » Sat Nov 09, 2019 11:53 am

tvubpwcisla wrote:
Sat Nov 09, 2019 11:51 am
Keeping your expenses as low as possible will help you fluctuate your withdrawal rate between a range that you feel comfortable with. Where you might run into some trouble is if your expenses dictate what your withdrawal rate needs to be. During a market correction, it is best to be in a place where you have a solid amount of cash on hand matched together with low monthly expense obligations. The folks that did extremely well during previous downturns in the economy and financial markets where the ones in exactly that place.
Exactly. This is why, for instance, not having a mortgage or rent in retirement reduces retirees' sequence of returns risk. Doing so enables retirees to more easily reduce their spending if/when their portfolio has poor performance. To the extent that retirees' expenses are fixed, the sequence of returns risk that they are exposed to increases.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

visualguy
Posts: 1600
Joined: Thu Jan 30, 2014 1:32 am

Re: Will we be OK at 3% WR?

Post by visualguy » Sat Nov 09, 2019 12:16 pm

tvubpwcisla wrote:
Sat Nov 09, 2019 11:51 am
Keeping your expenses as low as possible will help you fluctuate your withdrawal rate between a range that you feel comfortable with. Where you might run into some trouble is if your expenses dictate what your withdrawal rate needs to be. During a market correction, it is best to be in a place where you have a solid amount of cash on hand matched together with low monthly expense obligations. The folks that did extremely well during previous downturns in the economy and financial markets where the ones in exactly that place.
Right, but then you need to live the lifestyle of low expenses, which is not exactly what everyone dreams about for their retirement.

It works for people wealthy-enough to still live well even when cutting a whole bunch of discretionary spending. Basically, a low WR still enables a nice standard of living plus you would add a bunch of discretionary that can be cut at times.

It also works for those who have a low base of expenses because they can live in a frugal bare-bones manner, and everything else is discretionary that can be added or cut at times.

It doesn't work so well for the rest of us - pretty high base expenses and not all that much that we consider discretionary. We need an SWR that actually works quite well.

Regardless, I think there's more risk associated with not predicting expenses correctly than being too optimistic about the SWR, so I worry more about that side of things.

User avatar
willthrill81
Posts: 13594
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Will we be OK at 3% WR?

Post by willthrill81 » Sat Nov 09, 2019 12:19 pm

visualguy wrote:
Sat Nov 09, 2019 12:16 pm
tvubpwcisla wrote:
Sat Nov 09, 2019 11:51 am
Keeping your expenses as low as possible will help you fluctuate your withdrawal rate between a range that you feel comfortable with. Where you might run into some trouble is if your expenses dictate what your withdrawal rate needs to be. During a market correction, it is best to be in a place where you have a solid amount of cash on hand matched together with low monthly expense obligations. The folks that did extremely well during previous downturns in the economy and financial markets where the ones in exactly that place.
Right, but then you need to live the lifestyle of low expenses, which is not exactly what everyone dreams about for their retirement.
Not necessarily. Considering that housing is most people's single largest expense and can be greatly reduced by owning one's home outright (i.e. no mortgage), that can go a long way to reduce one's fixed expenses.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

emoore
Posts: 557
Joined: Mon Mar 04, 2013 8:16 pm

Re: Will we be OK at 3% WR?

Post by emoore » Sat Nov 09, 2019 12:34 pm

Thesaints wrote:
Fri Nov 08, 2019 6:39 pm
emoore wrote:
Fri Nov 08, 2019 6:35 pm
There is no evidence that this will happen.
It is simple math. Well, maybe not "simple" as in "everyone understands it easily", but more like "everyone with a solid grasp of statistics will have no problem agreeing".

Past data is only useful if one understands the conditions that generated it. Otherwise one could simply look at the history of the Chinese stock market returns and conclude that it will always be zero, as indeed it was for more than half a century.
I disagree that it's simple math. If it was there would be a lot more people able to time the market but they can't. So you can do whatever you would like but I'll continue to use historical data because it's better than guessing at the future.

am
Posts: 3041
Joined: Sun Sep 30, 2007 9:55 am

Re: Will we be OK at 3% WR?

Post by am » Sat Nov 09, 2019 12:58 pm

All these long posts over pages boil down to: 4% worked most of the time in the past but we are not sure about the future :D

User avatar
Will do good
Posts: 870
Joined: Fri Feb 24, 2012 8:23 pm

Re: Will we be OK at 3% WR?

Post by Will do good » Sat Nov 09, 2019 1:26 pm

I realized you can't use data or logic to debate this 4% discussion, those that insist 3% is reckless and 2% is the only safe choice will never agreed, no matter what is shown. For most of us the debate clearly shown 4% still the gold standard.

For those that willing to work much longer or willing to live way below their means, good luck.

TitaniumCranium
Posts: 27
Joined: Tue Oct 02, 2018 1:49 pm

Re: Will we be OK at 3% WR?

Post by TitaniumCranium » Sat Nov 09, 2019 3:18 pm

For whatever it's worth - my wife and I will retire in our mid 50s and plan on a 4% withdrawal rate ***BUT*** we plan on using not the Trinity study method of withdrawals but a constant % of current portfolio. 4% will allow us to upgrade our lifestyle with discretionary spending that will be approximately 61% of total expenses. When the markets go up YOY, we'll get a raise, when the markets go down YOY, we'll cut back on our discretionary spending. Cutting back when necessary will hopefully be painless. I think this shifts the sequence of return risk from completely running out of money before our demise, to decreased annual spending, which - given high discretionary spending - would be more tolerable.

I don't know that this offers anything of value to the debate - I just thought I'd mention this since I think everyone is focused on initial X% with annual inflation increases regardless of market performance.

visualguy
Posts: 1600
Joined: Thu Jan 30, 2014 1:32 am

Re: Will we be OK at 3% WR?

Post by visualguy » Sat Nov 09, 2019 3:32 pm

TitaniumCranium wrote:
Sat Nov 09, 2019 3:18 pm
For whatever it's worth - my wife and I will retire in our mid 50s and plan on a 4% withdrawal rate ***BUT*** we plan on using not the Trinity study method of withdrawals but a constant % of current portfolio. 4% will allow us to upgrade our lifestyle with discretionary spending that will be approximately 61% of total expenses. When the markets go up YOY, we'll get a raise, when the markets go down YOY, we'll cut back on our discretionary spending. Cutting back when necessary will hopefully be painless. I think this shifts the sequence of return risk from completely running out of money before our demise, to decreased annual spending, which - given high discretionary spending - would be more tolerable.

I don't know that this offers anything of value to the debate - I just thought I'd mention this since I think everyone is focused on initial X% with annual inflation increases regardless of market performance.
This works if you're at 61% discretionary for your expenses, but I would imagine that this is not common.

User avatar
Stef
Posts: 35
Joined: Thu Oct 10, 2019 10:13 am

Re: Will we be OK at 3% WR?

Post by Stef » Sat Nov 09, 2019 3:59 pm

TitaniumCranium wrote:
Sat Nov 09, 2019 3:18 pm
For whatever it's worth - my wife and I will retire in our mid 50s and plan on a 4% withdrawal rate ***BUT*** we plan on using not the Trinity study method of withdrawals but a constant % of current portfolio. 4% will allow us to upgrade our lifestyle with discretionary spending that will be approximately 61% of total expenses. When the markets go up YOY, we'll get a raise, when the markets go down YOY, we'll cut back on our discretionary spending. Cutting back when necessary will hopefully be painless. I think this shifts the sequence of return risk from completely running out of money before our demise, to decreased annual spending, which - given high discretionary spending - would be more tolerable.

I don't know that this offers anything of value to the debate - I just thought I'd mention this since I think everyone is focused on initial X% with annual inflation increases regardless of market performance.
If you adjust the withdrawals to the portfolio situation, couldn't you eben bump it up to 5%?

TitaniumCranium
Posts: 27
Joined: Tue Oct 02, 2018 1:49 pm

Re: Will we be OK at 3% WR?

Post by TitaniumCranium » Sat Nov 09, 2019 6:10 pm

Stef wrote:
Sat Nov 09, 2019 3:59 pm
If you adjust the withdrawals to the portfolio situation, couldn't you eben bump it up to 5%?
Sorry, I'm afraid I don't know the answer to that, nor am I confident enough to hazard a guess. My wife and I are good with 4% so...

Maybe someone else can answer that.

marcopolo
Posts: 2406
Joined: Sat Dec 03, 2016 10:22 am

Re: Will we be OK at 3% WR?

Post by marcopolo » Sat Nov 09, 2019 6:51 pm

Stef wrote:
Sat Nov 09, 2019 3:59 pm
TitaniumCranium wrote:
Sat Nov 09, 2019 3:18 pm
For whatever it's worth - my wife and I will retire in our mid 50s and plan on a 4% withdrawal rate ***BUT*** we plan on using not the Trinity study method of withdrawals but a constant % of current portfolio. 4% will allow us to upgrade our lifestyle with discretionary spending that will be approximately 61% of total expenses. When the markets go up YOY, we'll get a raise, when the markets go down YOY, we'll cut back on our discretionary spending. Cutting back when necessary will hopefully be painless. I think this shifts the sequence of return risk from completely running out of money before our demise, to decreased annual spending, which - given high discretionary spending - would be more tolerable.

I don't know that this offers anything of value to the debate - I just thought I'd mention this since I think everyone is focused on initial X% with annual inflation increases regardless of market performance.
If you adjust the withdrawals to the portfolio situation, couldn't you eben bump it up to 5%?
If you are withdrawing a percentage of remaining portfolio each year, then you will never run put of money, but your money available to spend will fluctuate, maybe quite bit. Theoretically you could use whatever percentage you want and never run out of money. The higher percentage you use, the more likely your withdrawal will someday be less money than you want to love on in a given year.
Once in a while you get shown the light, in the strangest of places if you look at it right.

Trader Joe
Posts: 1184
Joined: Fri Apr 25, 2014 6:38 pm

Re: Will we be OK at 3% WR?

Post by Trader Joe » Sat Nov 09, 2019 6:54 pm

bhsince87 wrote:
Tue Nov 05, 2019 11:23 pm
What's your opinion? Ages 54 and 52.

Will 3% be OK going forward?

Throw it all out there. I'd like to hear both sides.

Currently living on interest and dividends, which is about 2.2%

Are we sacrificing lifestyle enjoyment for paranoia, or a just a sense of safety?
Of course the answer is no. 3% is not enough and is not recommended.

marcopolo
Posts: 2406
Joined: Sat Dec 03, 2016 10:22 am

Re: Will we be OK at 3% WR?

Post by marcopolo » Sat Nov 09, 2019 7:22 pm

Trader Joe wrote:
Sat Nov 09, 2019 6:54 pm
bhsince87 wrote:
Tue Nov 05, 2019 11:23 pm
What's your opinion? Ages 54 and 52.

Will 3% be OK going forward?

Throw it all out there. I'd like to hear both sides.

Currently living on interest and dividends, which is about 2.2%

Are we sacrificing lifestyle enjoyment for paranoia, or a just a sense of safety?
Of course the answer is no. 3% is not enough and is not recommended.
What would you recommend? Why?
Once in a while you get shown the light, in the strangest of places if you look at it right.

User avatar
willthrill81
Posts: 13594
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Will we be OK at 3% WR?

Post by willthrill81 » Sat Nov 09, 2019 8:26 pm

Trader Joe wrote:
Sat Nov 09, 2019 6:54 pm
bhsince87 wrote:
Tue Nov 05, 2019 11:23 pm
What's your opinion? Ages 54 and 52.

Will 3% be OK going forward?

Throw it all out there. I'd like to hear both sides.

Currently living on interest and dividends, which is about 2.2%

Are we sacrificing lifestyle enjoyment for paranoia, or a just a sense of safety?
Of course the answer is no. 3% is not enough and is not recommended.
Of the course the answer is that no one knows what the SWR in the future will be. But 3% has historically been close to the perpetual withdrawal rate for balanced portfolios. So to the extent that the future resembles the past, it seems safe enough.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

User avatar
Stef
Posts: 35
Joined: Thu Oct 10, 2019 10:13 am

Re: Will we be OK at 3% WR?

Post by Stef » Sun Nov 10, 2019 2:42 am

marcopolo wrote:
Sat Nov 09, 2019 6:51 pm
If you are withdrawing a percentage of remaining portfolio each year, then you will never run put of money, but your money available to spend will fluctuate, maybe quite bit. Theoretically you could use whatever percentage you want and never run out of money. The higher percentage you use, the more likely your withdrawal will someday be less money than you want to live on in a given year.
It should be still less than the expected returns.

marcopolo
Posts: 2406
Joined: Sat Dec 03, 2016 10:22 am

Re: Will we be OK at 3% WR?

Post by marcopolo » Sun Nov 10, 2019 3:13 am

Stef wrote:
Sun Nov 10, 2019 2:42 am
marcopolo wrote:
Sat Nov 09, 2019 6:51 pm
If you are withdrawing a percentage of remaining portfolio each year, then you will never run put of money, but your money available to spend will fluctuate, maybe quite bit. Theoretically you could use whatever percentage you want and never run out of money. The higher percentage you use, the more likely your withdrawal will someday be less money than you want to live on in a given year.
It should be still less than the expected returns.
That seems like a prudent approach.
Of course, if you do then get the expected returns, you are likely to die with a big pile of money. Maybe not a bad problem to have. Another reason why "time value of money" discussed above makes a lot of sense.
Once in a while you get shown the light, in the strangest of places if you look at it right.

Post Reply