Larry Swedroe: 3% is the new 4%

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
EnjoyIt
Posts: 8272
Joined: Sun Dec 29, 2013 7:06 pm

Re: Larry Swedroe: 3% is the new 4%

Post by EnjoyIt »

Turbo29 wrote: Mon Feb 04, 2019 10:05 am
warowits wrote: Mon Feb 04, 2019 9:14 am
randomguy wrote: Mon Feb 04, 2019 8:12 am
naha66 wrote: Mon Feb 04, 2019 8:00 am
No Its the same, its just as easy for me to cut back because I'm use to living on less. Let's say I like to play golf 5 times a month and one of those rounds at a high end course. I just cut back to 3 times with no high end course. Just as someone with 5 million has a 50K a year golf club membership that he drops and then joins a cheaper club. I'm guessing you don't know how families live toward the lower end of the income spectrum.
What happens when you need to spend 100k to renovate your house due to some health issues that cause mobility issues? What about if you need to cut your housing costs in half? Easy to go from 4k ft to 2k. Going from 1200 to 600 is harder. Or your food budget? Giving up eating out is one thing. Skipping meals is another Going from a benz to a civic is easy. Going from a civic to a bike is harder😁
The idea that a person of regular means might have a mandatory unavoidable 100k house renovation actually made me laugh out loud. If that is how you use money I can see why retirement with only 500,000 plus social security would terrify you.
What if he had to replace an engine on his private jet the same time the house renovation came up?

Sometimes I think someone eating out every night has a harder time cutting back to cooking their own meals than someone cooking their own meals has cutting back by substituting chicken for beef.
Also, someone dropping spending by 10% from $50k/yr spending is dropping $5k. That can easily be that years vacation, going out a little less, cooking different foods, etc... Dropping 10% on a $200k/yr spending is $20k/yr which can be a substantially bigger difference.

Someone who wants to cut spending by $5k in one year can get a few credit card deals or a checking account opening deal and be half way there. a new credit card will do very little to curb $20k.

Maybe the $50k/yr person has it a little easier.
A time to EVALUATE your jitters: | viewtopic.php?p=1139732#p1139732
smitcat
Posts: 13300
Joined: Mon Nov 07, 2016 9:51 am

Re: Larry Swedroe: 3% is the new 4%

Post by smitcat »

EnjoyIt wrote: Mon Feb 04, 2019 5:47 pm
Turbo29 wrote: Mon Feb 04, 2019 10:05 am
warowits wrote: Mon Feb 04, 2019 9:14 am
randomguy wrote: Mon Feb 04, 2019 8:12 am
naha66 wrote: Mon Feb 04, 2019 8:00 am
No Its the same, its just as easy for me to cut back because I'm use to living on less. Let's say I like to play golf 5 times a month and one of those rounds at a high end course. I just cut back to 3 times with no high end course. Just as someone with 5 million has a 50K a year golf club membership that he drops and then joins a cheaper club. I'm guessing you don't know how families live toward the lower end of the income spectrum.
What happens when you need to spend 100k to renovate your house due to some health issues that cause mobility issues? What about if you need to cut your housing costs in half? Easy to go from 4k ft to 2k. Going from 1200 to 600 is harder. Or your food budget? Giving up eating out is one thing. Skipping meals is another Going from a benz to a civic is easy. Going from a civic to a bike is harder😁
The idea that a person of regular means might have a mandatory unavoidable 100k house renovation actually made me laugh out loud. If that is how you use money I can see why retirement with only 500,000 plus social security would terrify you.
What if he had to replace an engine on his private jet the same time the house renovation came up?

Sometimes I think someone eating out every night has a harder time cutting back to cooking their own meals than someone cooking their own meals has cutting back by substituting chicken for beef.
Also, someone dropping spending by 10% from $50k/yr spending is dropping $5k. That can easily be that years vacation, going out a little less, cooking different foods, etc... Dropping 10% on a $200k/yr spending is $20k/yr which can be a substantially bigger difference.

Someone who wants to cut spending by $5k in one year can get a few credit card deals or a checking account opening deal and be half way there. a new credit card will do very little to curb $20k.

Maybe the $50k/yr person has it a little easier.
Not likely...
Leesbro63
Posts: 10634
Joined: Mon Nov 08, 2010 3:36 pm

Re: Larry Swedroe: 3% is the new 4%

Post by Leesbro63 »

Socal77 wrote: Sun Feb 03, 2019 8:34 pm Except at 70 RMD is currently 3.65%.
What does RMD have to do with SWR? RMD is merely a taxation rule.
Socal77
Posts: 288
Joined: Thu Oct 19, 2017 1:14 pm

Re: Larry Swedroe: 3% is the new 4%

Post by Socal77 »

Leesbro63 wrote: Mon Feb 04, 2019 5:56 pm
Socal77 wrote: Sun Feb 03, 2019 8:34 pm Except at 70 RMD is currently 3.65%.
What does RMD have to do with SWR? RMD is merely a taxation rule.
So, if you have a 401K with $1,000,000 balance at age 70.5 you are not required to take a distribution of $36,500?

Maybe I am misunderstanding RMD's but if the above is true then the FED actuaries are saying 3.65% is a safe withdrawal rate for those aged 70.5+.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: 3% is the new 4%

Post by willthrill81 »

abc132 wrote: Mon Feb 04, 2019 12:06 pm
willthrill81 wrote: Mon Feb 04, 2019 11:57 am
When Bengen 'discovered' the '4% rule of thumb', it was the mere byproduct of certain specified parameters (e.g. fixed nominal spending plus annual inflation adjustment for a given AA). It was never intended to be a complete strategy for portfolio withdrawals, and virtually no one is using it that way. It is a historically conservative starting point, nothing more.
I'm sure that there are people that think they can safely withdraw 4% of their retirement amount.
I don't think that there are more than a handful that believe it's prudent to withdraw the same (or even more) from their portfolio when it's down significantly, not enough to be concerned about.
abc132 wrote: Mon Feb 04, 2019 12:06 pmIt serves very little basis as a starting point for withdrawals, because the amount you can withdraw can vary by a factor of 2 with identical portfolios.
Please expound upon this statement.
The Sensible Steward
User avatar
siamond
Posts: 6008
Joined: Mon May 28, 2012 5:50 am

Re: Larry Swedroe: 3% is the new 4%

Post by siamond »

Leesbro63 wrote: Mon Feb 04, 2019 5:56 pm
Socal77 wrote: Sun Feb 03, 2019 8:34 pm Except at 70 RMD is currently 3.65%.
What does RMD have to do with SWR? RMD is merely a taxation rule.
From the IRS standpoint, sure, it is primarily about taxes. But the way RMDs are computed is actually based on a very simple time value of money formula (aka PMT), something close to the Bogleheads VPW withdrawal method, except that the expected rate of return is zero. It is therefore not entirely crazy to just follow the IRS distribution as a withdrawal/budgeting method if all of your savings are in a (taxable) 401k/IRA. I mean, there are more optimal methods, but there are also much worse ones (notably the fixed -inflation-adjusted- withdrawal implied by SWR reasonings).

Here is a good article from Wade Pfau explaining how this works (this got me curious, I checked the math, this is indeed correct):
https://www.forbes.com/sites/wadepfau/2 ... a3a8c363ba
User avatar
Ben Mathew
Posts: 2720
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle

Re: Larry Swedroe: 3% is the new 4%

Post by Ben Mathew »

nedsaid wrote: Sun Feb 03, 2019 12:30 pm
metalworking wrote: Sun Feb 03, 2019 12:03 pm
nedsaid wrote: Sun Feb 03, 2019 11:45 am I will take one more crack at this and if Larry Swedroe thinks I am a blithering idiot, so be it. Something changed in American business after the 2008-2009 financial crisis and bear market. As I have interviewed for jobs, it has occurred to me that though labor is still very valuable, that it is less valuable than before. Artificial intelligence and automation have changed things, how much I am not sure. But I can see it as I have talked to employers, the world has changed. Expectations are so much different now than just 10-11 years ago. One reason that profit margins remain high are productivity improvements that we aren't perceiving yet.
The problem with this is you are just basing your beliefs (Swedroe model is wrong) on an as of yet undeveloped model (your thinking that something has changed).
Well, sometimes things really are different. The world changed with the industrial revolution. The world changed with the railroads. The world changed with air travel. The world changed with electrification. The world changed with the computer and information revolution. The world changed with the internet. Yes, I suppose stocks are "expensive" compared to 1860 but a lot has changed since then.
Productivity improvements need not be captured by individual firms and translated into profit. They can be competed away in the form of lower prices or increased costs. In the long run, it will be. Warren Buffett gave the example of airlines, and said something along the lines of stock investors would have been better off if the Wright brothers had been shot out of the sky.
The worst sort of business is one that grows rapidly, requires significant capital to engender the growth, and then earns little or no money. Think airlines. Here a durable competitive advantage has proven elusive ever since the days of the Wright Brothers. Indeed, if a farsighted capitalist had been present at Kitty Hawk, he would have done his successors a huge favor by shooting Orville down.
-- Warren Buffett, in the 2007 Berkshire Hathaway shareholder letter
The airline industry has expanded greatly and benefited from many technological improvements. It just doesn't translate to profits. It's the inverse of Paul Simon's lament in Can't Run But
The music suffers. The music business thrives.
Buffett's lament is:
The airline business suffers. The airline thrives.
Last edited by Ben Mathew on Mon Feb 04, 2019 11:16 pm, edited 1 time in total.
Total Portfolio Allocation and Withdrawal (TPAW)
User avatar
siamond
Posts: 6008
Joined: Mon May 28, 2012 5:50 am

Re: Larry Swedroe: 3% is the new 4%

Post by siamond »

marcopolo wrote: Mon Feb 04, 2019 10:33 amJust like the field has defined the idea of risk-adjusted returns, I think the academic researchers need to develop the idea of a risk-adjusted SWR, where the risk being considered is the probability of running out of money prior to death (joint for spouses), rather than for a fixed duration.
I don't have the pointer handy, but I believe Javier Estrada did exactly that, and you're right, this would make a lot of sense. Prof. Estrada is one of those very few retirement researchers actually producing novel research material every now and then - instead of rehashing endless SWR debates...
Socal77
Posts: 288
Joined: Thu Oct 19, 2017 1:14 pm

Re: Larry Swedroe: 3% is the new 4%

Post by Socal77 »

siamond wrote: Mon Feb 04, 2019 6:53 pm
Leesbro63 wrote: Mon Feb 04, 2019 5:56 pm
Socal77 wrote: Sun Feb 03, 2019 8:34 pm Except at 70 RMD is currently 3.65%.
What does RMD have to do with SWR? RMD is merely a taxation rule.
From the IRS standpoint, sure, it is primarily about taxes. But the way RMDs are computed is actually based on a very simple time value of money formula (aka PMT), something close to the Bogleheads VPW withdrawal method, except that the expected rate of return is zero. It is therefore not entirely crazy to just follow the IRS distribution as a withdrawal/budgeting method if all of your savings are in a (taxable) 401k/IRA. I mean, there are more optimal methods, but there are also much worse ones (notably the fixed -inflation-adjusted- withdrawal implied by SWR reasonings).

Here is a good article from Wade Pfau explaining how this works (this got me curious, I checked the math, this is indeed correct):
https://www.forbes.com/sites/wadepfau/2 ... a3a8c363ba
Thanks. This is what I thought and why I posted the current RMD for 70.5 plus is 3.65%. So at that point anything less is a moot point for SWR analysis.

But it does have value from the perspective that it provides a point of comparison for other withdrawal rates before you turn 70.5
randomguy
Posts: 11295
Joined: Wed Sep 17, 2014 9:00 am

Re: Larry Swedroe: 3% is the new 4%

Post by randomguy »

Socal77 wrote: Mon Feb 04, 2019 7:11 pm

Thanks. This is what I thought and why I posted the current RMD for 70.5 plus is 3.65%. So at that point anything less is a moot point for SWR analysis.

But it does have value from the perspective that it provides a point of comparison for other withdrawal rates before you turn 70.5
RMDs have nothing to do with SWR. They tell you that you have to take X out of an account. They dont say the payout will increase with inflation or last 30 years.
Socal77
Posts: 288
Joined: Thu Oct 19, 2017 1:14 pm

Re: Larry Swedroe: 3% is the new 4%

Post by Socal77 »

randomguy wrote: Mon Feb 04, 2019 7:26 pm
Socal77 wrote: Mon Feb 04, 2019 7:11 pm

Thanks. This is what I thought and why I posted the current RMD for 70.5 plus is 3.65%. So at that point anything less is a moot point for SWR analysis.

But it does have value from the perspective that it provides a point of comparison for other withdrawal rates before you turn 70.5
RMDs have nothing to do with SWR. They tell you that you have to take X out of an account. They dont say the payout will increase with inflation or last 30 years.
I understand. What I am attempting to communicate is the X they are telling you to take out, from their perspective, is an amount that will not completely deplete your account before you die.

They are saying that 3.65% is a safe withdrawal rate "floor" for someone 70.5+.

The inflation adjustment is irrelevant to what I am trying to communicate.
User avatar
siamond
Posts: 6008
Joined: Mon May 28, 2012 5:50 am

Re: Larry Swedroe: 3% is the new 4%

Post by siamond »

Socal77 wrote: Mon Feb 04, 2019 7:33 pmWhat I am attempting to communicate is the X they are telling you to take out, from their perspective, is an amount that will not completely deplete your account before you die.

They are saying that 3.65% is a safe withdrawal rate "floor" for someone 70.5+.
Well, yes and no. It is a percentage of your current portfolio at this point (say $1M), that would be roughly $36,500. Now let's say there is a truly major crisis, and your portfolio drops to $600k the year after. You're 71.5 by then, the IRS mandates that a distribution of 3.77% of your current portfolio (for that year), which would be roughly $22,600. And well, $22,600 is only 2.26% of the $1M you had to start with.

It is tricky to compare the SWR math (which is anchored on a percentage of the starting portfolio) with PMT-like withdrawal methods which are essentially percentages of the current portfolio.
Socal77
Posts: 288
Joined: Thu Oct 19, 2017 1:14 pm

Re: Larry Swedroe: 3% is the new 4%

Post by Socal77 »

siamond wrote: Mon Feb 04, 2019 8:20 pm
Socal77 wrote: Mon Feb 04, 2019 7:33 pmWhat I am attempting to communicate is the X they are telling you to take out, from their perspective, is an amount that will not completely deplete your account before you die.

They are saying that 3.65% is a safe withdrawal rate "floor" for someone 70.5+.
Well, yes and no. It is a percentage of your current portfolio at this point (say $1M), that is roughly $36500. This being said, let's say there is a truly major crisis, and your portfolio drops to $600k the year after. You're 71.5 by then, the IRS mandates that a distribution of 3.77% of your current portfolio (for that year), which would be roughly $22,600. And well, $22,600 is only 2.26% of the $1M you had to start with.

It is tricky to compare the SWR math (which is anchored on a percentage of the starting portfolio) with PMT-like withdrawal methods which are essentially percentages of the current portfolio.
That's a great way to communicate the subtleties of SWR being anchored at a starting portfolio sum vs. a distribution from a current portfolio sum but I'm not sure it makes a difference in what I am communicating - RMD's being a reference for thinking about how much to withdraw from a portfolio.

To make this even more simple lets just say you retire at 70.5 and are looking at what your SWR should be. Well, 3.65% is a good place to start because you're legally obligated to withdraw that amount.

I'm not saying you're wrong, you helped clarify your and other posters thinking that SWR are not related directly to RMD's.

I'm just saying the RMD gives you a sort of reference point for SWR's. I'm not saying they are the same. My thinking just helps me understand how "far off" some peoples SWR opinions may be.

I really fail to see how others can't see the logic in using RMD's as a reference point for the SWR, even though SWR is not RMD. Most people probably adjust their opinion of their personal withdrawal rate and SWR every year anyway according to market performance. After 70.5, the govt calcs your minimum SWR regardless of your original portfolio balance when you retired.
User avatar
siamond
Posts: 6008
Joined: Mon May 28, 2012 5:50 am

Re: Larry Swedroe: 3% is the new 4%

Post by siamond »

Socal77 wrote: Mon Feb 04, 2019 8:52 pmThat's a great way to communicate the subtleties of SWR being anchored at a starting portfolio sum vs. a distribution from a current portfolio sum but I'm not sure it makes a difference in what I am communicating - RMD's being a reference for thinking about how much to withdraw from a portfolio.

To make this even more simple lets just say you retire at 70.5 and are looking at what your SWR should be. Well, 3.65% is a good place to start because you're legally obligated to withdraw that amount.

I'm not saying you're wrong, you helped clarify your and other posters thinking that SWR are not related directly to RMD's.

I'm just saying the RMD gives you a sort of reference point for SWR's. I'm not saying they are the same. My thinking just helps me understand how "far off" some peoples SWR opinions may be.

I really fail to see how others can't see the logic in using RMD's as a reference point for the SWR, even though SWR is not RMD. Most people probably adjust their opinion of their personal SWR every year anyway according to market performance anyway. After 70.5, the govt calcs your SWR regardless of your original balance when you retired.
I think it's just a slight disconnect about terminology. You seem to be using 'SWR' as a withdrawal rate, to be applied to the current portfolio balance, and which would provide a fairly good safety (i.e. not withdrawing too much) in terms of your portfolio continuing to deliver decent (albeit variable) income during the rest of your retirement. And yes, the RMD rate does match this definition in a reasonable manner.

It's a fine way to define things, but when most people (notably researchers) speak of the SWR metric, they really mean the idea of history-tested (inflation-adjusted) fixed withdrawals. Where the rate only applies to the starting portfolio. This is not the same thing, even if the high-level idea is similar, and this is what I was trying to clarify.

But hey, no point getting hung up on terminology... As long as you're clear in your mind about what you're referring to, this is fine. And if you ask me, the RMD idea is WAY more sound than the fixed withdrawal approach, because it WILL do a decent job irrespective of what the future might throw at us, while the SWR approach might royally fail. This being said, I would encourage you to further investigate other time value of money approaches, the RMD approach is overly conservative and will skew the withdrawals upwards (on average, in the midst of the ups and downs of the market), which doesn't quite seem desirable. I actually recently wrote a blog article about such topic, take a look...
randomguy
Posts: 11295
Joined: Wed Sep 17, 2014 9:00 am

Re: Larry Swedroe: 3% is the new 4%

Post by randomguy »

Socal77 wrote: Mon Feb 04, 2019 7:33 pm
randomguy wrote: Mon Feb 04, 2019 7:26 pm
Socal77 wrote: Mon Feb 04, 2019 7:11 pm

Thanks. This is what I thought and why I posted the current RMD for 70.5 plus is 3.65%. So at that point anything less is a moot point for SWR analysis.

But it does have value from the perspective that it provides a point of comparison for other withdrawal rates before you turn 70.5
RMDs have nothing to do with SWR. They tell you that you have to take X out of an account. They dont say the payout will increase with inflation or last 30 years.
I understand. What I am attempting to communicate is the X they are telling you to take out, from their perspective, is an amount that will not completely deplete your account before you die.

They are saying that 3.65% is a safe withdrawal rate "floor" for someone 70.5+.

The inflation adjustment is irrelevant to what I am trying to communicate.
You can take out 90% of the account every year and you will not deplete the account before you die.:) The IRS isn't saying anything about SWR for anyone. It is a number that meets the tax needs of the US government, not the retirement needs of the account holders. There is no point in mixing the two together.

Now if you believe that the SWR is 3%, as other people have pointed out things like SPIA (something like 3.5-4% SWR) or just shoving all your money in tips (1% real gets you above 3.5%) gives better results. But if you go down either path you are giving up the chance for the 5-6%+ SWR (either more spending or big account on death) that will still happen 50%+ of the time. Remember it isn't like most people in the next 50 years will be facing 3% SWR. It will be one set of people that happens to retire just as a big bear market starts and end up with 15 years of 0% real return in stocks and bonds.
randomguy
Posts: 11295
Joined: Wed Sep 17, 2014 9:00 am

Re: Larry Swedroe: 3% is the new 4%

Post by randomguy »

EnjoyIt wrote: Mon Feb 04, 2019 5:47 pm
Also, someone dropping spending by 10% from $50k/yr spending is dropping $5k. That can easily be that years vacation, going out a little less, cooking different foods, etc... Dropping 10% on a $200k/yr spending is $20k/yr which can be a substantially bigger difference.

The 200k person just skips out on their 15k vacation and the 5k of taxes that it costs. How is that any harder?:)
User avatar
HomerJ
Posts: 21281
Joined: Fri Jun 06, 2008 12:50 pm

Re: Larry Swedroe: 3% is the new 4%

Post by HomerJ »

randomguy wrote: Mon Feb 04, 2019 9:46 pm
EnjoyIt wrote: Mon Feb 04, 2019 5:47 pm
Also, someone dropping spending by 10% from $50k/yr spending is dropping $5k. That can easily be that years vacation, going out a little less, cooking different foods, etc... Dropping 10% on a $200k/yr spending is $20k/yr which can be a substantially bigger difference.

The 200k person just skips out on their 15k vacation and the 5k of taxes that it costs. How is that any harder?:)
Yeah, and he still takes the other two 15k vacations he was planning that year. This is a silly argument.

Yes, the $50k guy can cut if needed. It's easier for the 200k guy.
"The best tools available to us are shovels, not scalpels. Don't get carried away." - vanBogle59
JackoC
Posts: 4714
Joined: Sun Aug 12, 2018 11:14 am

Re: Larry Swedroe: 3% is the new 4%

Post by JackoC »

nisiprius wrote: Mon Feb 04, 2019 7:18 am As a matter of curiosity, how does this relate to the Vanguard Managed Payout Fund, VGPDX?

It uses an actively managed portfolio with a modern, factor-aware, alternatives-aware allocation, and a 55% stock allocation which is two or three times more than that of their other "retirement income" funds (Wellesley, 38%; LifeStrategy Income, 20%; Target Retirement income, 30%).

1. So it has an "emphasis" on achieving an annual payout of 4% of capital while keeping pace with inflation--and "preserving capital over the long term," i.e. not spending down capital except temporarily. Past descriptions have made it clear that "preserving capital" means preserving it in real terms, although I don't see this explicitly here. The goal of "preserving capital" is clearly a higher one than merely "not going to zero."

2. Given that Vanguard itself has issued cautions about lower market returns going forward, doesn't this seem overoptimistic?
1. Yes it still says that explicitly in the latest prospectus
"Investment Objective
The Fund will make monthly cash distributions while seeking to have these
distributions and the invested capital keep pace with inflation over time."

2. It is at least somewhat at odds with Vanguard's own return expectations as per the "Vanguard Economic and Market Outlook" papers of recent years. The 2019 one lists the median projected 10 yr return of 60/40 'global balanced' portfolio as 4.9%, clearly stated as nominal not inflation adjusted, as of Sep 2018, up slightly from 4.5% in the previous year's version. so 2-3% real. For what little it's worth that's the kind of number I would use pre tax but tax is not a an asterisk for me: I figure around 2% real after tax for planning purposes. I don't suppose the difference between global balanced 60/40 stock/bond and 55 stock/45 'mix of various stuff' makes a dramatic difference.
https://pressroom.vanguard.com/nonindex ... 120618.pdf
https://institutional.vanguard.com/VGAp ... MktSummary

But it seems sometimes people sometimes see a '2 handle' discussed and get upset without fully considering the difference between expecting not to spend any inflation adjusted principal and being reasonably sure of principal not going below zero over 3 decades or so. If the realized return turns out to be 2% (which can't be ruled out even if you think it's low as midpoint expectation*) 4% (pre tax) of initial subsequently adjusted for inflation still takes ~35 yrs to run out. Not that I think 4% classic 'SWR' is conservative as some have termed it here. It's no longer IMO, not because market prices aren't 'justified'. But because they imply lower expected returns than historical and as has been correctly stated that's just a number which usually worked historically.

*the highest pre tax real return you can (almost) gtee is the long term TIPS real yield, around 1%.
zwzhang
Posts: 83
Joined: Thu Jul 25, 2013 10:43 am

Re: Larry Swedroe: 3% is the new 4%

Post by zwzhang »

willthrill81 wrote: Sun Feb 03, 2019 3:59 pm
larryswedroe wrote: Sun Feb 03, 2019 11:10 amSecond, it is exactly because we don't know that is the reason to be conservative, because the consequences of being wrong are literally unthinkable, while the alternatives are more acceptable.
Larry,

First of all, thanks for providing your reasoning for your statement.

I've heard this kind of statement made by critics of the '4% rule' before, but it's just plain silly. People often love to point out that if you strictly followed the '4% rule' in historical instances, you would generally have left a lot of money on the table unnecessarily or, in a few instances, essentially gone broke right after 30 years of withdrawals. But it's vital to keep in mind that no one is strictly following the '4% rule'. I've seriously never even heard of a rumor of anyone who did it or even seriously attempted to do so. Virtually everyone makes adjustments to their withdrawals, ratcheting them up when their portfolio does well and cutting them when the inverse occurs. No sane person is going to blindly spend down their portfolio to zero. The '4% rule' is used by most people around here at least in the way that they should use it: as an approximate guideline for how large a 'typical' (i.e. ~65 year old) retiree needs to produce a given amount of withdrawals.

Another problem with this notion is that it ignores the very important impact of the absolute size of one's portfolio. The '4% rule' is a lot more safe for someone with a $5 million portfolio than someone with a $500k portfolio, for instance, simply because it's very likely to be easier for someone starting with $5 million to reduce their spending and still have 'enough' than for someone starting with $500k.
Totally agree.
azanon
Posts: 3142
Joined: Mon Nov 07, 2011 9:34 am

Re: Larry Swedroe: 3% is the new 4%

Post by azanon »

international001 wrote: Mon Feb 04, 2019 5:31 pm SPIAs, SPIAs, SPIAs

Why nobody is advocating for them? I'm confused. Everybody wants to leave $$ to heirs?
Because a 28/72 portfolio (minimum variance) can beat the return of a typical inflation-adjusted SPIA, even if you don't count the sustained principle of the 28/72. Watch youtubes by "the Annuity Slayer", if you want to be left with no good reason to get a SPIA.
User avatar
siamond
Posts: 6008
Joined: Mon May 28, 2012 5:50 am

Re: Larry Swedroe: 3% is the new 4%

Post by siamond »

larryswedroe wrote: Sun Feb 03, 2019 11:10 am First, using valuations and yields is not an assumption, it's math. And we do know that valuations and yields are the BEST (though not great) metrics for estimating future returns. There is consensus on that in the academic literature, with about equal abililty in current valuations and CAPE metrics.
Using historical returns literally makes no sense unless the current valuations are about historic values. Perfect example is forecasting in 2000 when historical stock returns were in excess of 11% vs. say the 6% they were 70 years ago. That huge gap was to a large extent caused by rising valuations. And of course today real bond yields about half the historical average. And current yield curve are best estimate of future returns we have.
Er, not quite. 1/CAPE did display some (weak) predictive ability over a full retirement period when computing expected returns, this can be shown (I found very hard to believe it, but I ran the numbers, and there proved indeed true, at least with the known US data and as a WEAK signal). But that is absolutely NOT true for bond yields. Bond yields displayed a pretty good predictive ability for the following decade, but this deteriorates very fast (and even more in real terms, which is what matters here) and becomes plain noise when it comes to expected returns over 30 years, i.e. a typical retirement period. I don't like using historical returns for bonds any more than you do, but I am afraid that there is nothing else remotely valid that we can do.
larryswedroe wrote: Sun Feb 03, 2019 11:10 amSecond, it is exactly because we don't know that is the reason to be conservative, because the consequences of being wrong are literally unthinkable, while the alternatives are more acceptable. Note that most of the investment risk occurs in the first five years, but then as I explained much of the spending risk occurs later [...]
Sigh. Those various statements (dire consequences of being wrong, risk concentrated in first 5 years) are only true in the context of the utterly flawed fixed withdrawal method (X% of starting portfolio, keep it the same inflation-adjusted whatever happens in real life). It is quite really troublesome to see those misleading views be repeated ad nauseam by influential people. Please, oh please, get more familiar with variable withdrawal methods and their dynamics (e.g. Guyton-Klinger, Actuarial methods like VPW, etc) and you'll understand that one can take much better advantage of any future, dire or rosy, when using a sound retirement plan. Don't gate yourself by a known/unknown worst case, plan instead to be adaptive...
randomguy
Posts: 11295
Joined: Wed Sep 17, 2014 9:00 am

Re: Larry Swedroe: 3% is the new 4%

Post by randomguy »

azanon wrote: Tue Feb 05, 2019 10:25 am
international001 wrote: Mon Feb 04, 2019 5:31 pm SPIAs, SPIAs, SPIAs

Why nobody is advocating for them? I'm confused. Everybody wants to leave $$ to heirs?
Because a 28/72 portfolio (minimum variance) can beat the return of a typical inflation-adjusted SPIA, even if you don't count the sustained principle of the 28/72. Watch youtubes by "the Annuity Slayer", if you want to be left with no good reason to get a SPIA.
The portfolio wins about 95% of the time (depends a bit on your exact assumptions). The SPIA wins the other cases. It means you need to be very risk adverse to go the SPIA route at 65 for a good chunk of your money.
azanon
Posts: 3142
Joined: Mon Nov 07, 2011 9:34 am

Re: Larry Swedroe: 3% is the new 4%

Post by azanon »

randomguy wrote: Tue Feb 05, 2019 1:56 pm
azanon wrote: Tue Feb 05, 2019 10:25 am
international001 wrote: Mon Feb 04, 2019 5:31 pm SPIAs, SPIAs, SPIAs

Why nobody is advocating for them? I'm confused. Everybody wants to leave $$ to heirs?
Because a 28/72 portfolio (minimum variance) can beat the return of a typical inflation-adjusted SPIA, even if you don't count the sustained principle of the 28/72. Watch youtubes by "the Annuity Slayer", if you want to be left with no good reason to get a SPIA.
The portfolio wins about 95% of the time (depends a bit on your exact assumptions). The SPIA wins the other cases. It means you need to be very risk adverse to go the SPIA route at 65 for a good chunk of your money.
Yeah I thought it was a competitive product until I watched those youtubes I referenced, incidentally just this past weekend. He doesn't just give verbal opinions, rather goes through various excel spreadsheet comparisons, and shows just how ridiculous it would be to choose the SPIA. I used to believe SPIAs were the good annuities, but now I'm in the camp that none of them are good products, at least in comparison to low risk alternatives like a 30/70 using a reasonable withdrawal strategy.

So back on point, it's better to come up with a portfolio withdrawal strategy than use a SPIA.
visualguy
Posts: 2988
Joined: Thu Jan 30, 2014 12:32 am

Re: Larry Swedroe: 3% is the new 4%

Post by visualguy »

randomguy wrote: Tue Feb 05, 2019 1:56 pm
azanon wrote: Tue Feb 05, 2019 10:25 am
international001 wrote: Mon Feb 04, 2019 5:31 pm SPIAs, SPIAs, SPIAs

Why nobody is advocating for them? I'm confused. Everybody wants to leave $$ to heirs?
Because a 28/72 portfolio (minimum variance) can beat the return of a typical inflation-adjusted SPIA, even if you don't count the sustained principle of the 28/72. Watch youtubes by "the Annuity Slayer", if you want to be left with no good reason to get a SPIA.
The portfolio wins about 95% of the time (depends a bit on your exact assumptions). The SPIA wins the other cases. It means you need to be very risk adverse to go the SPIA route at 65 for a good chunk of your money.
It is past performance, though. I think people like to diversify among stock/bonds, SPIAs, and direct real estate because the future may be different.
azanon
Posts: 3142
Joined: Mon Nov 07, 2011 9:34 am

Re: Larry Swedroe: 3% is the new 4%

Post by azanon »

visualguy wrote: Tue Feb 05, 2019 3:29 pm
randomguy wrote: Tue Feb 05, 2019 1:56 pm
azanon wrote: Tue Feb 05, 2019 10:25 am
international001 wrote: Mon Feb 04, 2019 5:31 pm SPIAs, SPIAs, SPIAs

Why nobody is advocating for them? I'm confused. Everybody wants to leave $$ to heirs?
Because a 28/72 portfolio (minimum variance) can beat the return of a typical inflation-adjusted SPIA, even if you don't count the sustained principle of the 28/72. Watch youtubes by "the Annuity Slayer", if you want to be left with no good reason to get a SPIA.
The portfolio wins about 95% of the time (depends a bit on your exact assumptions). The SPIA wins the other cases. It means you need to be very risk adverse to go the SPIA route at 65 for a good chunk of your money.
It is past performance, though. I think people like to diversify among stock/bonds, SPIAs, and direct real estate because the future may be different.
I was mainly just answering the original question - were we just worried about leaving money to our heirs? The answer is, no we're also worried about getting a worse return on investment than the expected return of the most conservative risk-adjusted return portfolio that can be constructed, so that's why most of us avoid SPIAs. The only obvious upside of a SPIA is for individual or company selling them. They must be huge money-makers.
JackoC
Posts: 4714
Joined: Sun Aug 12, 2018 11:14 am

Re: Larry Swedroe: 3% is the new 4%

Post by JackoC »

siamond wrote: Tue Feb 05, 2019 10:56 am
larryswedroe wrote: Sun Feb 03, 2019 11:10 am First, using valuations and yields is not an assumption, it's math. And we do know that valuations and yields are the BEST (though not great) metrics for estimating future returns. There is consensus on that in the academic literature, with about equal abililty in current valuations and CAPE metrics.
Using historical returns literally makes no sense unless the current valuations are about historic values. Perfect example is forecasting in 2000 when historical stock returns were in excess of 11% vs. say the 6% they were 70 years ago. That huge gap was to a large extent caused by rising valuations. And of course today real bond yields about half the historical average. And current yield curve are best estimate of future returns we have.
Er, not quite. 1/CAPE did display some (weak) predictive ability over a full retirement period when computing expected returns, this can be shown (I found very hard to believe it, but I ran the numbers, and there proved indeed true, at least with the known US data and as a WEAK signal). But that is absolutely NOT true for bond yields. Bond yields displayed a pretty good predictive ability for the following decade, but this deteriorates very fast (and even more in real terms, which is what matters here) and becomes plain noise when it comes to expected returns over 30 years, i.e. a typical retirement period. I don't like using historical returns for bonds any more than you do, but I am afraid that there is nothing else remotely valid that we can do
Not quite as clearly with stocks because of their inherently greater uncertainty, but with bonds I believe Larry Swedroe is correct and you are mistaken. It is irrelevant what bonds returned in the past compared to what they yield today for any investment horizon for which there are bonds today. So, what will bonds return over the next 100 yrs? You can't see from the yield curve. But for 30 yrs you either chose today's real return (around 1% for TIPS) or else take more risk to inflation (30 yr nominal), or else take more rollover risk (5 or 10 yr TIPS or nominal and roll over). The higher risk alternatives might turn out with higher realized return, but there is no reason to think the *expected* return would be higher doing that. In fact one secondary relevant fact from past bond returns is the term premium was generally positive (you made more locking in longest rates than shorter ones and rolling over**). So Larry S is basically 100% right there, far from there being no alternative to estimating expected return on bonds by looking at historical bonds returns, there is no reason to do that. It's pretty clearly incorrect to do that.

Again with stocks everything is fuzzier. But, there is an underlying price/yield relationship like that of bonds. The current PE or cyclically adjusted PE might not have a high correlation with the future *realized* return, because realized returns over moderate periods (people often look at a decade) are heavily influenced by changes in valuation, the so called speculative return. But there's still a fundamental reason to look at it to estimate the expected return. As Larry S says, a fundamental relationship between PE and expected return via the dividend discount model. It's not like hypothesizing 'maybe stock returns depend if the NFC or AFC wins the Super Bowl that year' and then arguing how the historical statistics support this or not. And on an expected basis there's no reason to think valuation will increase (some people think it should have an expected *decrease* because of 'reversion to the mean', but E[r] based on today's valuations is well under historical even without assuming that). IOW periods where realized return exceeded 1/PE[x] due to valuation increases are not really relevant to *expected* return now. Nobody says realized return couldn't be higher than expected, that should always be something like a 50-50 (not exactly if you want to get technical) shot. But Larry S is also basically correct in pointing to the big flaw in assuming expected return for stocks equals past historical realized return without considering valuation *at all*.

*there are some corporate and non-US govt bonds that long, but 'bonds' generally means US govt bonds here.
**some term structure models now say the term premium is negative, very unusual, there was a debate about it on a recent thread, but it's still a matter of 10's of bps. It's ridiculous to estimate today's expected return on long term bonds over the next 30 yrs as the long term past historical geometric average, around 3% real, when the 30yr TIPS yield is 1% real, and the 30 yr nominal is 3%. And bond E[r] is 2% lower than historical, clearly and obviously, but stock E[r] is not lower at all?
User avatar
Hyperborea
Posts: 956
Joined: Sat Apr 15, 2017 10:31 am
Location: Portugal

Re: Larry Swedroe: 3% is the new 4%

Post by Hyperborea »

HomerJ wrote: Mon Feb 04, 2019 12:54 pm Plus, no one here recommends 100% stock the day before you retire. The 4% plan assumes a 60/40 portfolio, so you won't drop 40% in one year.
The "4% plan" does not "assume" a 60/40 portfolio. The research showed that there is a broad plateau of equity allocations that have worked historically. For a 30 year retirement horizon (i.e. a 65-70 year old retiree) that plateau is from 50% to 100% equity. If things go ahistorically wrong then it's likely the ends of the plateau that are in danger so maybe the middle is the safest putting the range from about 65% to 85%.

For younger retirees in their 50's particularly for those who are married, not even the really young ones in their 30's, they need to look at higher equity allocations to cover a potentially 40 year plus retirement.
It’s not just that facts don’t seem to matter anymore. It’s that it doesn’t seem to matter that facts don’t matter.
abc132
Posts: 2435
Joined: Thu Oct 18, 2018 1:11 am

Re: Larry Swedroe: 3% is the new 4%

Post by abc132 »

willthrill81 wrote: Mon Feb 04, 2019 6:42 pm
abc132 wrote: Mon Feb 04, 2019 12:06 pm
willthrill81 wrote: Mon Feb 04, 2019 11:57 am
When Bengen 'discovered' the '4% rule of thumb', it was the mere byproduct of certain specified parameters (e.g. fixed nominal spending plus annual inflation adjustment for a given AA). It was never intended to be a complete strategy for portfolio withdrawals, and virtually no one is using it that way. It is a historically conservative starting point, nothing more.
I'm sure that there are people that think they can safely withdraw 4% of their retirement amount.
I don't think that there are more than a handful that believe it's prudent to withdraw the same (or even more) from their portfolio when it's down significantly, not enough to be concerned about.
abc132 wrote: Mon Feb 04, 2019 12:06 pmIt serves very little basis as a starting point for withdrawals, because the amount you can withdraw can vary by a factor of 2 with identical portfolios.
Please expound upon this statement.
I already gave the examples. Person A retires with 2 million and withdraws 80,000 per year. After a 40% market loss, Person B with an equivalent portfolio retires after the loss, and now has 1.2 million and can only withdraw 48,000 per year. It doesn't matter what the portfolio distribution is, both people with identical portfolios suffer the same loss.

Those amounts are so vastly different as to make a 4% withdrawal plan or guide worthless.

Larry is telling you this by saying you might need to plan for 3% based on a more complex model and Monte Carlo simulations, which is exactly what Person A might need to do. Person B could get by just fine with 4% or even 5%, because of better valuations and better future expectations at the time they retire.

Note that Person A with 3% steady withdrawal rate and person B with 5% steady withdrawal rate both spend 60,000 per year with the same 1.2 million dollars.

People will adjust their spending (if needed) to some degree, but the reality is that only extended periods (10+ years) of reduced spending really have much impact, and spending less in a downturn does very little to your success chance, because portfolio losses (~10x expenses in this example) are much bigger than any reduced spending (0.2x expenses).
JackoC
Posts: 4714
Joined: Sun Aug 12, 2018 11:14 am

Re: Larry Swedroe: 3% is the new 4%

Post by JackoC »

azanon wrote: Tue Feb 05, 2019 4:08 pm
visualguy wrote: Tue Feb 05, 2019 3:29 pm
randomguy wrote: Tue Feb 05, 2019 1:56 pm
azanon wrote: Tue Feb 05, 2019 10:25 am
international001 wrote: Mon Feb 04, 2019 5:31 pm SPIAs, SPIAs, SPIAs

Why nobody is advocating for them? I'm confused. Everybody wants to leave $$ to heirs?
Because a 28/72 portfolio (minimum variance) can beat the return of a typical inflation-adjusted SPIA, even if you don't count the sustained principle of the 28/72. Watch youtubes by "the Annuity Slayer", if you want to be left with no good reason to get a SPIA.
The portfolio wins about 95% of the time (depends a bit on your exact assumptions). The SPIA wins the other cases. It means you need to be very risk adverse to go the SPIA route at 65 for a good chunk of your money.
It is past performance, though. I think people like to diversify among stock/bonds, SPIAs, and direct real estate because the future may be different.
I was mainly just answering the original question - were we just worried about leaving money to our heirs? The answer is, no we're also worried about getting a worse return on investment than the expected return of the most conservative risk-adjusted return portfolio that can be constructed, so that's why most of us avoid SPIAs. The only obvious upside of a SPIA is for individual or company selling them. They must be huge money-makers.
But you're ignoring the point about past performance of the 28/72 being compared to annuity rates now. That's a big flaw in the 'Annuity Slayer' analysis, at least on the couple of video's I looked up, one mentioning the 28/72 historical results. That's the problem Larry Swedroe is talking about, which is particularly acute for portfolios with lots of bonds. The long term historical return on long term US bonds (back to early 20th century) was in the ballpark of 3% real. The current real expected return of long term bonds is only around 1% (1% TIPS yield or 3% nominal 30 yr yield if inflation expectation is 2%). That gives a big artificial advantage comparing 28/72 historical returns to SPIA's priced off today's yield curve.

That part of the AS pitch seems like BS, to be blunt. Some of the other points they make, despite the annoying tone, are more valid. Such as *IF* you buy an SPIA from an adviser who piles on fees. But you can also buy one through Vanguard for example and get around that.

SPIA's are not 'the' answer, but the idea that they are beaten by conservative stock/bond portolio's virtually all the time is based on two fallacies
1) as just explained, comparing SPIA's priced off a 1% bond real return yield curve now to past performance of heavily bond portfolios when realized real bond returns were 2% points higher than that. But long term bonds you buy *now* will have the same expected return as the ones the ins co buys now to back the SPIA, around 1% real not 3% real.
2) more obviously, the SPIA will look worthless if you assume you know how long you're going to live and it's not extremely long. But you don't know, which is the whole idea of annuities. I didn't see all the AS video's but in the couple I made it through they didn't even acknowledge that difference. Annuities cover risk of living much longer than you expect to, though imperfectly for various reasons. A simple collection of ETF's can only address that risk by lowering the withdrawal rate.
User avatar
siamond
Posts: 6008
Joined: Mon May 28, 2012 5:50 am

Re: Larry Swedroe: 3% is the new 4%

Post by siamond »

JackoC wrote: Tue Feb 05, 2019 4:22 pm
siamond wrote: Tue Feb 05, 2019 10:56 am
larryswedroe wrote: Sun Feb 03, 2019 11:10 am First, using valuations and yields is not an assumption, it's math. And we do know that valuations and yields are the BEST (though not great) metrics for estimating future returns. [...]
Er, not quite. 1/CAPE did display some (weak) predictive ability over a full retirement period when computing expected returns, this can be shown (I found very hard to believe it, but I ran the numbers, and there proved indeed true, at least with the known US data and as a WEAK signal). But that is absolutely NOT true for bond yields. Bond yields displayed a pretty good predictive ability for the following decade, but this deteriorates very fast (and even more in real terms, which is what matters here) and becomes plain noise when it comes to expected returns over 30 years, i.e. a typical retirement period. I don't like using historical returns for bonds any more than you do, but I am afraid that there is nothing else remotely valid that we can do
Not quite as clearly with stocks because of their inherently greater uncertainty, but with bonds I believe Larry Swedroe is correct and you are mistaken. [...]
Well, I didn't make a statement out of the blue... I ran the numbers, many times and published the results on this forum in a large expected returns spreadsheet a while ago. Bonds (and interest rates) are totally unpredictable on the long run, and even more in real terms. Yes, there are good solid reasons for bond yields to be a solid model for the following decade, but unfortunately, this doesn't hold muster on the longer term because of the interest rates unpredictability. I am actually quite frustrated by the finding myself, but this is just a plain fact. Don't take my word (or Larry's), please take the data from multpl.com and run the numbers by themselves... Now are historical averages any better here? Not by much, I give you that. There is just no sensible alternative I know of.
JackoC wrote: Tue Feb 05, 2019 4:22 pmStocks everything is fuzzier. But, there is an underlying price/yield relationship like that of bonds. The current PE or cyclically adjusted PE might not have a high correlation with the future *realized* return, because realized returns over moderate periods (people often look at a decade) are heavily influenced by changes in valuation, the so called speculative return. But there's still a fundamental reason to look at it to estimate the expected return. As Larry S says, a fundamental relationship between PE and expected return via the dividend discount model. [...] But Larry S is also basically correct in pointing to the big flaw in assuming expected return for stocks equals past historical realized return without considering valuation *at all*.
I actually don't disagree with you here. If you read my post again, I did acknowledge that 1/CAPE displayed a weak predictive ability over 20 to 30 years (which is to say that starting valuations do matter, although certainly not that much over such a long period), at least with US data. And yes, you're right, one can do some algebra along the lines of the DDM model to get a sense of why that is. It does remain a weak signal though. Which is again easy to verify by running the numbers yourself.
JackoC
Posts: 4714
Joined: Sun Aug 12, 2018 11:14 am

Re: Larry Swedroe: 3% is the new 4%

Post by JackoC »

siamond wrote: Tue Feb 05, 2019 5:28 pm
JackoC wrote: Tue Feb 05, 2019 4:22 pm Not quite as clearly with stocks because of their inherently greater uncertainty, but with bonds I believe Larry Swedroe is correct and you are mistaken. [...]
1. Well, I didn't make a statement out of the blue... I ran the numbers, many times and published the results on this forum in a large expected returns spreadsheet a while ago.
JackoC wrote: Tue Feb 05, 2019 4:22 pmStocks everything is fuzzier. But, there is an underlying price/yield relationship like that of bonds. The current PE or cyclically adjusted PE might not have a high correlation with the future *realized* return, because realized returns over moderate periods (people often look at a decade) are heavily influenced by changes in valuation, the so called speculative return. But there's still a fundamental reason to look at it to estimate the expected return. As Larry S says, a fundamental relationship between PE and expected return via the dividend discount model. [...] But Larry S is also basically correct in pointing to the big flaw in assuming expected return for stocks equals past historical realized return without considering valuation *at all*.
I actually don't disagree with you here. If you read my post again, I did acknowledge that 1/CAPE displayed a weak predictive ability over 20 to 30 years (which is to say that starting valuations do matter, although certainly not that much over such a long period), at least with US data. And yes, you're right, one can do some algebra along the lines of the DDM model to get a sense of why that is. It does remain a weak signal though. Which is again easy to verify by running the numbers yourself.
1. I'm not arguing whether you ran the numbers of what past bond returns are. I'm simply stating that Larry S is correct in stating that past bond returns are almost entirely irrelevant to bond expected returns. We know what the *real expected return* of bonds for the next 30 yrs is, 1% as per the TIPS yield or 3% nominal 30 yr yield given a 2% expected inflation rate. Simple as that. We don't know the realized return from investing in 5yr TIPS and rolling them over 6 times. We don't know the realized real return of the 30 nominal, depending on realized inflation. But there is no reason to choose as *expected* return a number other than long term TIPS yield, or long term nominal minus *expected* inflation.

2. I'm glad we don't entirely disagree. But again the noise of past returns which accounts for 'weakness' isn't necessarily relevant to expected return, just as past bond returns are almost entirely irrelevant, just not to the same degree with stocks, because they contain so many more uncertainties. Or more specifically that's the reason 1/PE[x] (and choosing x) is fuzzier estimate of stock E[r] than long term bond yields are of bond E[r]. What stocks returned in the past is still not directly relevant to what they will return in the future, without considering differences in the basic supply/demand equation for capital, and the risk environment. IOW estimating E[r] w/o considering valuation is not correct. It's just more, and more obviously, incorrect when it comes to govt bonds.
User avatar
Ben Mathew
Posts: 2720
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle

Re: Larry Swedroe: 3% is the new 4%

Post by Ben Mathew »

JackoC wrote: Tue Feb 05, 2019 5:38 pm
siamond wrote: Tue Feb 05, 2019 5:28 pm
JackoC wrote: Tue Feb 05, 2019 4:22 pm Not quite as clearly with stocks because of their inherently greater uncertainty, but with bonds I believe Larry Swedroe is correct and you are mistaken. [...]
1. Well, I didn't make a statement out of the blue... I ran the numbers, many times and published the results on this forum in a large expected returns spreadsheet a while ago.
JackoC wrote: Tue Feb 05, 2019 4:22 pmStocks everything is fuzzier. But, there is an underlying price/yield relationship like that of bonds. The current PE or cyclically adjusted PE might not have a high correlation with the future *realized* return, because realized returns over moderate periods (people often look at a decade) are heavily influenced by changes in valuation, the so called speculative return. But there's still a fundamental reason to look at it to estimate the expected return. As Larry S says, a fundamental relationship between PE and expected return via the dividend discount model. [...] But Larry S is also basically correct in pointing to the big flaw in assuming expected return for stocks equals past historical realized return without considering valuation *at all*.
I actually don't disagree with you here. If you read my post again, I did acknowledge that 1/CAPE displayed a weak predictive ability over 20 to 30 years (which is to say that starting valuations do matter, although certainly not that much over such a long period), at least with US data. And yes, you're right, one can do some algebra along the lines of the DDM model to get a sense of why that is. It does remain a weak signal though. Which is again easy to verify by running the numbers yourself.
1. I'm not arguing whether you ran the numbers of what past bond returns are. I'm simply stating that Larry S is correct in stating that past bond returns are almost entirely irrelevant to bond expected returns. We know what the *real expected return* of bonds for the next 30 yrs is, 1% as per the TIPS yield or 3% nominal 30 yr yield given a 2% expected inflation rate. Simple as that. We don't know the realized return from investing in 5yr TIPS and rolling them over 6 times. We don't know the realized real return of the 30 nominal, depending on realized inflation. But there is no reason to choose as *expected* return a number other than long term TIPS yield, or long term nominal minus *expected* inflation.

2. I'm glad we don't entirely disagree. But again the noise of past returns which accounts for 'weakness' isn't necessarily relevant to expected return, just as past bond returns are almost entirely irrelevant, just not to the same degree with stocks, because they contain so many more uncertainties. Or more specifically that's the reason 1/PE[x] (and choosing x) is fuzzier estimate of stock E[r] than long term bond yields are of bond E[r]. What stocks returned in the past is still not directly relevant to what they will return in the future, without considering differences in the basic supply/demand equation for capital, and the risk environment. IOW estimating E[r] w/o considering valuation is not correct. It's just more, and more obviously, incorrect when it comes to govt bonds.
The advantage to focusing on yields rather than past performance is that it helps you avoid getting drawn into bubbles. Past performance tends to make assets that have had a recent run-up in prices, and is therefore overpriced, look good precisely when they are the most expensive. Yields look at both sides of the equation: price in relation to value. That, IMO, makes it a more robust signal.

To address Siamond's empirical claim that yields have not that been that useful, I think the right comparison here is yields vs past performance. Does past performance predict future performance better than yields? I would be surprised if it did, especially if you're looking at past performance over horizons shorter than thirty years.

Ideally, I think expected return would condition on both current valuations and the long term performance (>30 years) of an asset class.
Total Portfolio Allocation and Withdrawal (TPAW)
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: 3% is the new 4%

Post by willthrill81 »

siamond wrote: Tue Feb 05, 2019 5:28 pm
JackoC wrote: Tue Feb 05, 2019 4:22 pm
siamond wrote: Tue Feb 05, 2019 10:56 am
larryswedroe wrote: Sun Feb 03, 2019 11:10 am First, using valuations and yields is not an assumption, it's math. And we do know that valuations and yields are the BEST (though not great) metrics for estimating future returns. [...]
Er, not quite. 1/CAPE did display some (weak) predictive ability over a full retirement period when computing expected returns, this can be shown (I found very hard to believe it, but I ran the numbers, and there proved indeed true, at least with the known US data and as a WEAK signal). But that is absolutely NOT true for bond yields. Bond yields displayed a pretty good predictive ability for the following decade, but this deteriorates very fast (and even more in real terms, which is what matters here) and becomes plain noise when it comes to expected returns over 30 years, i.e. a typical retirement period. I don't like using historical returns for bonds any more than you do, but I am afraid that there is nothing else remotely valid that we can do
Not quite as clearly with stocks because of their inherently greater uncertainty, but with bonds I believe Larry Swedroe is correct and you are mistaken. [...]
Well, I didn't make a statement out of the blue... I ran the numbers, many times and published the results on this forum in a large expected returns spreadsheet a while ago. Bonds (and interest rates) are totally unpredictable on the long run, and even more in real terms. Yes, there are good solid reasons for bond yields to be a solid model for the following decade, but unfortunately, this doesn't hold muster on the longer term because of the interest rates unpredictability. I am actually quite frustrated by the finding myself, but this is just a plain fact. Don't take my word (or Larry's), please take the data from multpl.com and run the numbers by themselves... Now are historical averages any better here? Not by much, I give you that. There is just no sensible alternative I know of.
JackoC wrote: Tue Feb 05, 2019 4:22 pmStocks everything is fuzzier. But, there is an underlying price/yield relationship like that of bonds. The current PE or cyclically adjusted PE might not have a high correlation with the future *realized* return, because realized returns over moderate periods (people often look at a decade) are heavily influenced by changes in valuation, the so called speculative return. But there's still a fundamental reason to look at it to estimate the expected return. As Larry S says, a fundamental relationship between PE and expected return via the dividend discount model. [...] But Larry S is also basically correct in pointing to the big flaw in assuming expected return for stocks equals past historical realized return without considering valuation *at all*.
I actually don't disagree with you here. If you read my post again, I did acknowledge that 1/CAPE displayed a weak predictive ability over 20 to 30 years (which is to say that starting valuations do matter, although certainly not that much over such a long period), at least with US data. And yes, you're right, one can do some algebra along the lines of the DDM model to get a sense of why that is. It does remain a weak signal though. Which is again easy to verify by running the numbers yourself.
+1

Just to add, Kitces noted years ago that in comparing the highest and lowest historic quintiles of CAPE, the difference in the subsequent 30 year returns was just +/- 1%. That's not nothing, but it's far from earth shattering either. It certainly wasn't enough to cast doubt on the '4% rule'.

Image
https://www.kitces.com/blog/should-equi ... valuation/

And yes, to the extent that the future resembles the past, current bond yields are a very weak predictor of long-term future returns.

I really don't understand why people get into such a dither about expected low returns from equities and bonds alone casting doubt on the '4% rule'. The '4% rule' was based around some pretty terrible sequences of poor returns and, if volatility weren't an issue, would succeed with a real portfolio return of just 1.22%.

Here's a thought that you would be good at analyzing: if a retiree interested in fixed nominal withdrawals (which very few retirees actually are) was really concerned that the '4% rule' wouldn't hold up over the next 30 years, couldn't they just go all-in to TIPS (e.g. mostly 5 and 10 year TIPS) and then slowly ramp up their equity exposure (i.e. a 'reverse' glidepath, but using TIPS with a guaranteed real return instead)? This would seemingly remove the worst of sequence of returns risk, the biggest threat to the success of the '4% rule', while also guaranteeing real returns over that first decade.
The Sensible Steward
Seasonal
Posts: 3045
Joined: Sun May 21, 2017 1:49 pm

Re: Larry Swedroe: 3% is the new 4%

Post by Seasonal »

willthrill81 wrote: Tue Feb 05, 2019 7:11 pmI really don't understand why people get into such a dither about expected low returns from equities and bonds alone casting doubt on the '4% rule'. The '4% rule' was based around some pretty terrible sequences of poor returns and, if volatility weren't an issue, would succeed with a real portfolio return of just 1.22%.
The reason I hear most often is that the combination of current p/e and current real bond yields results in lower expected returns than in the past, even if some past actual returns have been bad. I do not know if this is actually true.
willthrill81 wrote: Tue Feb 05, 2019 7:11 pmHere's a thought that you would be good at analyzing: if a retiree interested in fixed nominal withdrawals (which very few retirees actually are) was really concerned that the '4% rule' wouldn't hold up over the next 30 years, couldn't they just go all-in to TIPS (e.g. mostly 5 and 10 year TIPS) and then slowly ramp up their equity exposure (i.e. a 'reverse' glidepath, but using TIPS with a guaranteed real return instead)? This would seemingly remove the worst of sequence of returns risk, the biggest threat to the success of the '4% rule', while also guaranteeing real returns over that first decade.
Zvi Bodie recommends TIPS covering needed expenses and, once that's covered, additional amounts in equities for potential upside. For example, see https://www.forbes.com/sites/wadepfau/2 ... 92684f5c8e

An issue with using 5 and 10 year TIPS is reinvestment risk.
User avatar
HomerJ
Posts: 21281
Joined: Fri Jun 06, 2008 12:50 pm

Re: Larry Swedroe: 3% is the new 4%

Post by HomerJ »

Seasonal wrote: Tue Feb 05, 2019 8:14 pm
willthrill81 wrote: Tue Feb 05, 2019 7:11 pmI really don't understand why people get into such a dither about expected low returns from equities and bonds alone casting doubt on the '4% rule'. The '4% rule' was based around some pretty terrible sequences of poor returns and, if volatility weren't an issue, would succeed with a real portfolio return of just 1.22%.
The reason I hear most often is that the combination of current p/e and current real bond yields results in lower expected returns than in the past, even if some past actual returns have been bad. I do not know if this is actually true.
4% worked during the Great Depression.

For 4% to not work, stock and bond returns going forward would have to be WORSE than the Great Depression. You know, where stocks fell 89%?

Forecasting "lower expected returns" is one thing. Forecasting "The next 30 years will be the WORST 30 years in U.S. history" is an entirely different thing.

Lower expected returns because valuations are high shouldn't kill 4%. 4% is based on periods where we had lower expected returns.

People who retired in 2000 retired with higher valuations and even lower expected returns. And yet 4% is working for them.

People who retired in 1966 had 16 years of terrible stock returns AND terrible bond returns AND double-digit inflation. And yet 4% (okay 3.8%) worked for them.

You can get more than 3% guaranteed for life using SPIAs. You can get more than 3% guaranteed for 30 years using TIPs.
Last edited by HomerJ on Tue Feb 05, 2019 10:39 pm, edited 1 time in total.
"The best tools available to us are shovels, not scalpels. Don't get carried away." - vanBogle59
User avatar
HomerJ
Posts: 21281
Joined: Fri Jun 06, 2008 12:50 pm

Re: Larry Swedroe: 3% is the new 4%

Post by HomerJ »

A 60-year old male can get a 6% SPIA right now.

Put half your money in it, now you're getting 3% for life (not inflation-adjusted), and you still have half your money left, where the dividends alone will be at least 2% (and depending on your stock/bond mix, likely more than 2%), putting you at 4%+ withdrawals of your original starting amount.

Easy peasy lemon squeezy.
"The best tools available to us are shovels, not scalpels. Don't get carried away." - vanBogle59
randomguy
Posts: 11295
Joined: Wed Sep 17, 2014 9:00 am

Re: Larry Swedroe: 3% is the new 4%

Post by randomguy »

HomerJ wrote: Tue Feb 05, 2019 10:35 pm A 60-year old male can get a 6% SPIA right now.

Put half your money in it, now you're getting 3% for life (not inflation-adjusted), and you still have half your money left, where the dividends alone will be at least 2% (and depending on your stock/bond mix, likely more than 2%), putting you at 4%+ withdrawals of your original starting amount.

Easy peasy lemon squeezy.
Sure it is easy but it is also pretty darn horrible compared to the 4% SWR. Lets compare for the rate limiting step of 1966. They both start at 40k. Where are they at a mere 15 years later

20k SPIA->7k of purchasing power
20k of portfolio spending(doesn't really matter the split stocks and bonds had close to the same return and assuming you are spending divs/interest) is reduced to 10k due to portfolio shrinkage.

so great you have gone from 40k to 17k of spending in a 15 years. Does that sound like a remotely appealing retirement plan or would you rather get 40k for 28 years and then die broke?:) And yes that is very back of the envelope math but you can see how bad it is to invest in a nominal instrument when inflation shows up. Obviously that is about the worst possible year for this scheme and yes your payout in 1966 was probably higher given the starting interest rates. But even if you were getting 10%, having your money drop by 66% in value is brutal.
User avatar
HomerJ
Posts: 21281
Joined: Fri Jun 06, 2008 12:50 pm

Re: Larry Swedroe: 3% is the new 4%

Post by HomerJ »

randomguy wrote: Wed Feb 06, 2019 12:13 am
HomerJ wrote: Tue Feb 05, 2019 10:35 pm A 60-year old male can get a 6% SPIA right now.

Put half your money in it, now you're getting 3% for life (not inflation-adjusted), and you still have half your money left, where the dividends alone will be at least 2% (and depending on your stock/bond mix, likely more than 2%), putting you at 4%+ withdrawals of your original starting amount.

Easy peasy lemon squeezy.
Sure it is easy but it is also pretty darn horrible compared to the 4% SWR. Lets compare for the rate limiting step of 1966. They both start at 40k. Where are they at a mere 15 years later

20k SPIA->7k of purchasing power
20k of portfolio spending(doesn't really matter the split stocks and bonds had close to the same return and assuming you are spending divs/interest) is reduced to 10k due to portfolio shrinkage.

so great you have gone from 40k to 17k of spending in a 15 years. Does that sound like a remotely appealing retirement plan or would you rather get 40k for 28 years and then die broke?:) And yes that is very back of the envelope math but you can see how bad it is to invest in a nominal instrument when inflation shows up. Obviously that is about the worst possible year for this scheme and yes your payout in 1966 was probably higher given the starting interest rates. But even if you were getting 10%, having your money drop by 66% in value is brutal.
Heh you make good points, but I can still make a case.

First off, half your portfolio gives you 3%, and the other half gives you 1%. Not 2% and 2%.

$1 million dollars in 1966:
  • spend $500,000 on 6% SPIA (probably higher back then, but let's stick with it). - That gives you $30k a year.
  • invest $500,000, pull 2% a year. - That gives you $10k a year.
15 years later:
  • 30k SPIA is only worth 11k due to inflation. That is indeed bad.
  • But that $500,000 has grown... Dividends were so high back then, that you still made (nominal) money if you were only pulling 2%. Stock prices were flat from 1966 to 1981 (DOW was 1000 in 1966 and still 1000 in 1981, but they were throwing off 4%-5% dividends)
Now you're 75, interest rates are through the roof, and you still have a good chunk of money. You can probably get another annuity at 15% or more. (even today, 75-year-old males can get 9.2%).

You need to get your income back to the original $40k a year in 1966 dollars ($114,000 in 1981 dollars). You still have $30k a year coming in. In 1981 dollars, you'd need another $84,000.

But you've still got $700,000 or so, at 15% you'd need to annuitize $560,000 of it to get $84,000 and get yourself back to $114,000 a year income in 1981 (equal to $40,000 in 1966).

But yeah, that's cutting it close... :)

But the good news is you were probably a smoker in 1966, and you're already dead.
"The best tools available to us are shovels, not scalpels. Don't get carried away." - vanBogle59
politely
Posts: 144
Joined: Thu Jun 09, 2016 11:20 pm

Re: Larry Swedroe: 3% is the new 4%

Post by politely »

HomerJ wrote: Sun Feb 03, 2019 10:22 pm
Far better to retire at 4%, with the expectation of MAYBE going back to work (or cutting back some, but not all, fun) if bad things happen.

Isn't that better than working for 5 more years to make sure you don't have to go back to work if bad things happen. But you already worked 5 more years! (so technically you just guaranteed that bad things happened)
I think reasonable minds can differ on this point. I plan on having more than 30 years of retirement, and for me, the worst case scenario would be having to look for a new job when I'm in my 70s or 80s or hitting my kids up for an allowance. Maybe it's over-caution or simply pride, but I'd rather work a few more years now than potentially have to go through that scenario later, when I may not have the same job opportunities or my experience has become obsolete, and my salary will likely be lower. Again, I think it's subjective, but I am very willing to take on five more years today if it ensures (to the extent possible) that I won't have to go back to work for five more years in twenty or thirty years time.
stlutz
Posts: 5585
Joined: Fri Jan 02, 2009 12:08 am

Re: Larry Swedroe: 3% is the new 4%

Post by stlutz »

I think this thread has demonstrated that making 4% (or 3.85% or whatever) work is really quite easy given today's annuity or TIPS rates. The advantage that the annuity has over the bonds is the included longevity insurance; TIPS of course don't have [as much] credit risk. In short, as of today, interest rates are not the problem.

Equity returns are the problem.

The whole question of equity "expected returns" kind of misses the point as this issue is more risk than return. Again, "good" equity returns are not what's needed. The problem is a combination of a poor sequence of returns and poor overall returns. If you have both, 4% doesn't work on a heavily equity-weighted portfolio.

4% worked in 1966 because equities launched into a strong bull market in 1982. That didn't *have* to happen. When I read the infamous "Death of Equities" Business Week article from 1979, what sticks with me is not that the article was dumb (it wasn't). Rather, what sticks with me is that the problems identified in the article were solved or at least mitigated. Turning that non-required result into a portfolio withdrawal "rule" is problematic. Poor returns combined with a poor sequence of returns has a higher probability of happening than past US stock market performance might suggest.

My problem with the with the original analysis launching this thread and others like it is the underlying assumption that market pricing is incorrect. Low PE ratios suggest higher risk, not lower risk. How many people on this board were suggesting that March, 2009 was a good time to retire following the 4% rule? Not many! Was the board collectively wrong?

Based on a cold analysis of the facts at the time, I would say no. PE ratios were low at the time because the market was very risky then. Higher risk means a higher probability of low returns combined with a poor sequence of returns. Retiring in 2008/2009 is working out to be one of those times when a withdrawal rate much higher than 4% works. However, that good fortune came with a higher risk that this would be a period when 4% didn't work. Once again, the problems we had at the time were handled relatively well compared to the possible outcomes. That didn't have to happen.

But to step back from the market timing discussion, the fact is that stocks are always a risky investment. The person investing in stocks is *always* taking a risk of having poor returns (and a poor sequence of those returns). Reduce current corporate profit margins by half over the long haul (not an unrealistic assumption) and you have a recipe for poor future returns.

If one needs 4% and they really don't have the flexibility to have more variable spending, the answer is a combination of annuities and duration-matched fixed income (mostly TIPS). If you don't have the ability to take the equity risk, you shouldn't be doing so.
User avatar
siamond
Posts: 6008
Joined: Mon May 28, 2012 5:50 am

Re: Larry Swedroe: 3% is the new 4%

Post by siamond »

willthrill81 wrote: Tue Feb 05, 2019 7:11 pmI really don't understand why people get into such a dither about expected low returns from equities and bonds alone casting doubt on the '4% rule'. The '4% rule' was based around some pretty terrible sequences of poor returns and, if volatility weren't an issue, would succeed with a real portfolio return of just 1.22%.
I only discussed the 'expected returns' theme because this is the way Larry justified this '3% SWR' assertion. I totally agree with you (and others) that this makes no sense, a really bad SWR is much more about really bad sequences of return than a low CAGR, and we've seen unbalanced situations like today in the (US) past that did NOT produce SWR lower than 4%. Which means that Larry is mixing up timeframes (coming decade vs. full retirement period) in TWO ways. First, his computation of expected returns (squarely for bonds and partly for stocks) has little logical or empirical basis for a full retirement period. Next, his application of such result to an SWR prediction makes no sense either unless you think 'decade' instead of '30 years'. Personally, I only participate to such threads to try to steer people towards fact-based thinking, but it is really annoying to see influential people on this board peddling such non-sense.
willthrill81 wrote: Tue Feb 05, 2019 7:11 pmHere's a thought that you would be good at analyzing: if a retiree interested in fixed nominal withdrawals (which very few retirees actually are) was really concerned that the '4% rule' wouldn't hold up over the next 30 years, couldn't they just go all-in to TIPS (e.g. mostly 5 and 10 year TIPS) and then slowly ramp up their equity exposure (i.e. a 'reverse' glidepath, but using TIPS with a guaranteed real return instead)? This would seemingly remove the worst of sequence of returns risk, the biggest threat to the success of the '4% rule', while also guaranteeing real returns over that first decade.
Those of us who ran reverse glidepath numbers on this board found them unconvincing, and even the authors of such concept acknowledged in a roundabout way (in follow up papers) the evidence was flimsy. More importantly, this is *again* in a fixed withdrawal context, which nobody in their right mind should even consider. I just cannot see how this would help anybody with a sound (i.e. variable/adaptive) withdrawal plan, where SoR issues are largely mitigated. As to backtesting with TIPS, unfortunately, no, can't do. Researchers who tried to come up with TIPS models (plus our own attempt at it) didn't go anywhere. So we have just a couple of decades of TIPS history, and there is no way we can do any proper 'retirement period' analysis on those. Nice try though!
JackoC
Posts: 4714
Joined: Sun Aug 12, 2018 11:14 am

Re: Larry Swedroe: 3% is the new 4%

Post by JackoC »

willthrill81 wrote: Tue Feb 05, 2019 7:11 pm Just to add, Kitces noted years ago that in comparing the highest and lowest historic quintiles of CAPE, the difference in the subsequent 30 year returns was just +/- 1%. That's not nothing, but it's far from earth shattering either. It certainly wasn't enough to cast doubt on the '4% rule'.
But this is under the fallacious assumption that if there's lots of statistical noise when comparing past expected and past subsequent realized returns, then the default assumption should be that today's expected return is the past realized return.

Again in case of bonds that assumption is clearly nonsensical. If you have a 30 yr horizon today and want to invest in bonds the expected return is the 30 yr yield. It is not the historical realized return, obviously. You can instead invest to a 5 yr horizon and roll it over 6 times, you can instead buy a 30 yr bond today, sell it and buy another one next year and so on for 30 yrs. You can do all kinds of things which don't simply lock into today's available return, and those *might* come out ahead due to statistical randomness. But there is no logic in assuming the *expected* outcome, the centroid of the statistical distribution would be higher than the 30 yr yield. If the market is in any way efficient. If the 30 yr TIPS yield/30 yr nominal are 1%/3% with an apparently prevailing 2% inflation expectation, but rolling over 5 yr notes or buying selling the 30 yr every yr had an *expected* return 2% points higher (the long term historical realized real return was around 3%), then the 30 yr is grossly mispriced*. But if assume you can discern gross mispricings like that in liquid instruments, the whole rationale of BH'ism falls apart. It is nonsense, there is no less blunt way to put it, to assume realized historical bond returns as the expected return now in a 60/40 or whatever % bond portfolio you put together. The correct estimate is today's yields.*

Again this argument does not carry over 100% to stocks, but I think if it's fully understood it still casts a lot of doubt on the argument you made about past statistics. Siamond was correct to say you can assemble past statistics of a roll over or constant maturity strategy in bonds and find it's very noisy in realized return compared to the bond yields at the beginning of the period. But this proves exactly nothing in support of the idea we should use past realized bond returns as the expected return today. Likewise everyone is aware that past subsequent realized stocks returns were very noisy compared to stock expected return (as estimated by 1/PE[x]) at the beginning of the relevant period. That doesn't lend much logical support to the idea that now's expected return is the past realized return.

It ignores the basic fact that the supply/demand for capital and the risk environment can change. There's a reason bond expected returns are now so much lower than past realized bond returns, and it's not reasonable to assume that that reason** has no impact on stock expected returns v past realized. Just showing noisy graphs of the past expected/realize relationship doesn't justify that assumption.

* again as above give or take term premium, under a positive term premium as has generally prevailed the roll over would have lower return, the constant mat higher than locking in for the investment horizon, but now some models price a negative term premium.
**lower expected returns on capital (almost) without credit risk. Moreover the trend toward generally lower credit spreads in the risky bond market shows that risk isn't priced as highly now as it was in the past data set. All the more reason to doubt the assumption that the stock expected return now must be the average past realized return unless proved otherwise.
cheezit
Posts: 493
Joined: Sat Jul 14, 2018 7:28 pm

Re: Larry Swedroe: 3% is the new 4%

Post by cheezit »

siamond wrote: Wed Feb 06, 2019 6:04 am More importantly, this is *again* in a fixed withdrawal context, which nobody in their right mind should even consider.
Certainly, people have fixed costs of living, which are affected by inflation as time goes by. A variable withdrawal plan assumes that you have a substantial margin of non-essential spending baked into your costs that you can eat into when lean years come wrt. returns on your investments; for someone living frugally, this may not be true. The frugal retiree then has to have some other margin in case things don't go well, namely starting retirement with a greater ratio of savings to expenses. Going from a nest egg of 25x expenses to 33x expenses before starting retirement is just going from 4% to 3% by another name.


[A] really bad SWR is much more about really bad sequences of return than a low CAGR, and we've seen unbalanced situations like today in the (US) past that did NOT produce SWR lower than 4%.
If this were true, the observed SWR for other countries that had lower CAGR for equities than the US but mostly didn't crash harder would not be lower than the US rate of ~4%. In fact, the SWR for an international investor has been substantially lower than that of a US investor, around 3.2% if memory serves. If the expected return of assets goes 1% or 1.5% lower but the sequence of returns is equally "bumpy" going forward as it was looking back, the SWR for the forward-looking period *will* be lower than the SWR for the backward-looking period was.

Of course, the caveat here (besides the usual crystal ball cloudiness about expected returns in general, and the limitations/flaws of the model assuming a 30 year retirement) is, will the CAGR for the forward-looking period be lower than the average of the backward-looking period, or will it be lower than the worst 30-year slice of the backward-looking period? The latter needs to be the case for the SWR for the forward-looking period to go down since the 4% number is already based on the worst couple of slices of the entire backward-looking period.
abc132
Posts: 2435
Joined: Thu Oct 18, 2018 1:11 am

Re: Larry Swedroe: 3% is the new 4%

Post by abc132 »

stlutz wrote: Wed Feb 06, 2019 1:54 am I think this thread has demonstrated that making 4% (or 3.85% or whatever) work is really quite easy given today's annuity or TIPS rates. The advantage that the annuity has over the bonds is the included longevity insurance; TIPS of course don't have [as much] credit risk. In short, as of today, interest rates are not the problem.

Equity returns are the problem.

The whole question of equity "expected returns" kind of misses the point as this issue is more risk than return. Again, "good" equity returns are not what's needed. The problem is a combination of a poor sequence of returns and poor overall returns. If you have both, 4% doesn't work on a heavily equity-weighted portfolio.

4% worked in 1966 because equities launched into a strong bull market in 1982. That didn't *have* to happen. When I read the infamous "Death of Equities" Business Week article from 1979, what sticks with me is not that the article was dumb (it wasn't). Rather, what sticks with me is that the problems identified in the article were solved or at least mitigated. Turning that non-required result into a portfolio withdrawal "rule" is problematic. Poor returns combined with a poor sequence of returns has a higher probability of happening than past US stock market performance might suggest.
I agree with this, in that we should at least consider the possibility of poor timing and poor sequence of returns. There is nothing about the 4% rule that guarantees anything, it is just a historical guide. Maybe even a great one, but not a guarantee.
stlutz wrote: Wed Feb 06, 2019 1:54 am My problem with the with the original analysis launching this thread and others like it is the underlying assumption that market pricing is incorrect. Low PE ratios suggest higher risk, not lower risk. How many people on this board were suggesting that March, 2009 was a good time to retire following the 4% rule? Not many! Was the board collectively wrong?
Yes they were, and that is why the 4% rule is used so poorly. I have given the example of 2 people with identical portfolios, and this example shows that if the 4% rule was good enough before a market drop, something higher than 4% must be good enough for a new retiree after a market drop. This is a requirement of two people with equal portfolios that retire before and after a bad sequence of returns being able to withdraw the same amount of money per year with the same portfolio and having the same chance of success.

If you believe in the 4% rule at the top, you should believe in something greater than 4% (for a new retiree) at the bottom. Not doing so says that the person retiring at the top using the 4% rule is expected to fail, when the expected safety of the 4% rule is the entire reason it is being used as a guide.
randomguy
Posts: 11295
Joined: Wed Sep 17, 2014 9:00 am

Re: Larry Swedroe: 3% is the new 4%

Post by randomguy »

HomerJ wrote: Wed Feb 06, 2019 12:34 am

Heh you make good points, but I can still make a case.

First off, half your portfolio gives you 3%, and the other half gives you 1%. Not 2% and 2%.

$1 million dollars in 1966:
  • spend $500,000 on 6% SPIA (probably higher back then, but let's stick with it). - That gives you $30k a year.
  • invest $500,000, pull 2% a year. - That gives you $10k a year.
15 years later:
  • 30k SPIA is only worth 11k due to inflation. That is indeed bad.
  • But that $500,000 has grown... Dividends were so high back then, that you still made (nominal) money if you were only pulling 2%. Stock prices were flat from 1966 to 1981 (DOW was 1000 in 1966 and still 1000 in 1981, but they were throwing off 4%-5% dividends)
Now you're 75, interest rates are through the roof, and you still have a good chunk of money. You can probably get another annuity at 15% or more. (even today, 75-year-old males can get 9.2%).

You need to get your income back to the original $40k a year in 1966 dollars ($114,000 in 1981 dollars). You still have $30k a year coming in. In 1981 dollars, you'd need another $84,000.

But you've still got $700,000 or so, at 15% you'd need to annuitize $560,000 of it to get $84,000 and get yourself back to $114,000 a year income in 1981 (equal to $40,000 in 1966).

But yeah, that's cutting it close... :)

But the good news is you were probably a smoker in 1966, and you're already dead.
Sorry I screwed the annuity math up. Should not do math after midnight:)

Now we didn't define this problem well but lets see how your math works out. Guy hits 75 with 140k to pay for inflation adjustments on 114k of spending over the next decade. That might work out. But what were the years from 65-75 like? In 1980 you were getting 30k from the annuity and maybe 15k from the portfolio (2% of 750k or so) while you would have need like 100k to maintain your standard of living. If you take out the 55k to do that, you aren't ending up with 700k in 1981 or even 500k. Obviously we are half assing this whole analysis and you would need to get exact returns, spending, and SPIA purchase dollars. In the studies that I have seen, SPIAs don't increase SWR at all over 30 years. You just lock in getting the worst results in exchange for no market volatility.

And I am not sure smoking is a good retirement planning tool:)
fourkids
Posts: 210
Joined: Mon Mar 23, 2015 9:40 am

Re: Larry Swedroe: 3% is the new 4%

Post by fourkids »

why not 2% or 1%? where does it stop? This feels like it is just a marketing plot to make us keep more assets in our investment accounts that advisors can charge fees on.

Seriously, has anyone ever gone broke withdrawing 4%? My goal is not to work longer than I have to and die with a big pot of cash.
wolf359
Posts: 3207
Joined: Sun Mar 15, 2015 8:47 am

Re: Larry Swedroe: 3% is the new 4%

Post by wolf359 »

TheTimeLord wrote: Mon Feb 04, 2019 10:25 am
flyingaway wrote: Mon Feb 04, 2019 10:11 am I now have satisfied the 4% rule. It seems that we need to work on the 3% rule.
In a few years, I am afraid that someone will be talking about a 2% rule.
We can work until we are dead, which does not need any rule.
Even though I am a perennial member of the OMY club I am somewhere between terrified and flummoxed by these threads. Throw in concerns about LTC and you can start feeling like the concept of retirement is impractical for anyone without an 8 digit portfolio. Seriously, what are the odds of being alive after 30 years into retirement for most people?
The odds of an individual who is 65 years old living for another 30 years is low (about 25%, but depending upon whether male or female). But when measuring the life expectancy for couples who are 65, the odds jump to 50% for at least one of them living for another 25 years. (There are multiple sources for this data, including the CDC, the Social Security Administration, and I've seen it cited in studies by Swedrow and Kitces.)

If you retire earlier than 65, you should allow for a longer retirement period.
User avatar
HomerJ
Posts: 21281
Joined: Fri Jun 06, 2008 12:50 pm

Re: Larry Swedroe: 3% is the new 4%

Post by HomerJ »

abc132 wrote: Wed Feb 06, 2019 9:35 amI agree with this, in that we should at least consider the possibility of poor timing and poor sequence of returns. There is nothing about the 4% rule that guarantees anything, it is just a historical guide. Maybe even a great one, but not a guarantee.
Correct
I have given the example of 2 people with identical portfolios, and this example shows that if the 4% rule was good enough before a market drop, something higher than 4% must be good enough for a new retiree after a market drop.
Sure, but how much higher? Smart move is to just take the 4%, and AFTER the expected large gains appear, ratchet up your spending. That way you are still safe if the gains don't appear.

If I retire at 4% withdrawals, and my portfolio doubles, you better believe I'll increase my spending or giving a bit... Maybe not quite all the way up to 4% of the new portfolio value (although technically I could), but nothing says I have to stay at that bottom spending level forever.

4% or 25x expenses is just a GUIDE of when you probably have enough to retire.

Your contrived example doesn't make any sense in the real world.
(1) Bogleheads that close to retirement aren't 100% stocks
(2) Someone who could retire in 2009 with half their money gone would have retired in 2008 instead.
"The best tools available to us are shovels, not scalpels. Don't get carried away." - vanBogle59
User avatar
HomerJ
Posts: 21281
Joined: Fri Jun 06, 2008 12:50 pm

Re: Larry Swedroe: 3% is the new 4%

Post by HomerJ »

randomguy wrote: Wed Feb 06, 2019 9:47 amNow we didn't define this problem well but lets see how your math works out. Guy hits 75 with 140k to pay for inflation adjustments on 114k of spending over the next decade. That might work out. But what were the years from 65-75 like? In 1980 you were getting 30k from the annuity and maybe 15k from the portfolio (2% of 750k or so) while you would have need like 100k to maintain your standard of living. If you take out the 55k to do that, you aren't ending up with 700k in 1981 or even 500k. Obviously we are half assing this whole analysis and you would need to get exact returns, spending, and SPIA purchase dollars. In the studies that I have seen, SPIAs don't increase SWR at all over 30 years. You just lock in getting the worst results in exchange for no market volatility.

And I am not sure smoking is a good retirement planning tool:)
Yeah, you're right.. Runaway inflation could a huge killer for a SPIA plan.
"The best tools available to us are shovels, not scalpels. Don't get carried away." - vanBogle59
User avatar
HomerJ
Posts: 21281
Joined: Fri Jun 06, 2008 12:50 pm

Re: Larry Swedroe: 3% is the new 4%

Post by HomerJ »

fourkids wrote: Wed Feb 06, 2019 9:53 amSeriously, has anyone ever gone broke withdrawing 4%? My goal is not to work longer than I have to and die with a big pot of cash.
Not yet. (Well, technically, it failed starting in 1966 and 1967. You ran out of money in 27 or 28 years I think. 3.8% worked for the full 30 years though).

But again, that's if you withdrew the full 4%, inflation-adjusted each year, without any variance.

In the real world, most of us can cut back spending a bit.

For most of us here, "Failure" doesn't mean being broke. It means taking 2 vacations a year instead of 4 vacations a year for some period of time while waiting for the stock market to recover.

I could retire at 55, and have a 96% chance of taking 4 vacations a year through my entire retirement, and a 4% chance of only taking 2 vacations for a good chunk of those years.

Or... I could work an extra 5 long years, just to get my 4 vacations a year chances up to 100%.

No thanks. Especially since I've already seen friends and family DIE in their 50s and 60s.
"The best tools available to us are shovels, not scalpels. Don't get carried away." - vanBogle59
Post Reply