Valuation-dependent Safe Withdrawal Rate?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
getrichslowly
Posts: 73
Joined: Mon Feb 12, 2018 11:48 am

Valuation-dependent Safe Withdrawal Rate?

Post by getrichslowly » Mon Feb 12, 2018 11:52 am

Has anybody done any analysis on what the maximum Safe Withdrawal Rate is as a function of current valuations?

The intuition is that when valuations are high, expected future return is lower, so the maximum safe withdrawal rate should also be lower.

I don't think valuations can inform optimal asset allocation, but I do think they ought to inform optimal retirement point. If the traditional strategy supported by historical data is that one can retire as soon as you have 25x living expenses (i.e. 4% SWR) in savings, then this target may instead be a function of current valuations. e.g. maybe at current market valuations, one should wait until they have 30x, 35x, or possibly 40x living expenses in savings. I don't know. The purpose of this question is to find out.

If no one has conducted this analysis before, I can be the first to contribute an original analysis. Otherwise if someone else has already done it, I would rather not duplicate work.

cjg
Posts: 61
Joined: Sat Jan 11, 2014 9:03 pm

Re: Valuation-dependent Safe Withdrawal Rate?

Post by cjg » Mon Feb 12, 2018 1:36 pm

Some analysis here. In my opinion, we don't really have enough data to understand the effect very well.

https://earlyretirementnow.com/2016/12/ ... valuation/

james22
Posts: 1187
Joined: Tue Aug 21, 2007 2:22 pm

Re: Valuation-dependent Safe Withdrawal Rate?

Post by james22 » Mon Feb 12, 2018 2:09 pm

james22 wrote:
Wed Aug 16, 2017 5:24 am
Who Ate Joe’s Retirement Money? Sequence Risk and its Insidious Drag on Retirement Wealth

With the observations that starting valuation can be a useful tool in forecasting future returns and that owning stocks when they are cheaper tends to win in the long term with the additional benefit of suffering less damaging drawdowns, let’s conduct a simple experiment.

Suppose we use starting CAPEs as our simple and sole indicator of future stock market returns. Let’s go back in history as far as we can and pretend that we are managing the glidepath, but this time, we will manage it dynamically – that is, underweighting or overweighting equities and fixed income around the predetermined path, based on one simple valuation metric, which changes through time. Basically, if we believed stocks were expensive, we would underweight them, and if stocks appeared cheap, we would overweight them.

Can we do better than the static (predetermined) glidepath in terms of drawdowns and final wealth?

The answer is apparently yes.


https://www.gmo.com/docs/default-source ... ?sfvrsn=20

Investing for Retirement: The Defined Contribution Challenge

Let us assume we have a worker who turned 55 in 1965, and to that point was on target for retirement savings. The worker, however, had the misfortune of being in peak savings years during a period in which equity valuations were high and real bond yields were low. The period from the mid-1960s through the 1970s represents some of the worst real returns for both stocks and bonds on record.

A static, or inflexible, glide path would ignore these lofty valuations, and ensure the investor had a higher weighting in equities in 1965, when valuations were very high, than in 1974, when valuations were much lower. The fully dynamic stock weight is based on minimizing expected shortfall incorporating time-varying expected returns.

At first glance, the dynamic flight path looks nonsensical. It shows no weight in stocks in 1965, when the participant is 55 years old and has another 10 years until retirement, but by 1974 the stock weight rises to 90%, staying at a very high level until the early 1980s despite the fact that the participant is approaching 70 years old, and losses can be devastating.

But if the goal is minimizing expected shortfall of wealth in retirement, it can make sense to run an aggressive portfolio in retirement if the increase in expected returns is high enough. Furthermore, if the expected return to stocks is actually lower than bonds due to high valuations, such as was the case in 1965, it is hard to see why owning stocks would help at all.

The static strategy leaves the participant out of money by 1992, the cost of not taking into account the changing valuations of the stock and bond markets over time, and consuming a constant real dollar amount equal to 5% of target wealth. The dynamic strategy, by contrast, allows the participant to make up for earlier inadequate returns, and the money lasts for a full 30-year retirement even though consumption is high.


https://www.gmo.com/docs/default-source ... f?sfvrsn=0
This whole episode is likely to end so badly that future children will learn about it in school and shake their heads in wonder at the rank stupidity of it all... Hussman

User avatar
willthrill81
Posts: 4063
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Valuation-dependent Safe Withdrawal Rate?

Post by willthrill81 » Mon Feb 12, 2018 2:20 pm

getrichslowly wrote:
Mon Feb 12, 2018 11:52 am
Has anybody done any analysis on what the maximum Safe Withdrawal Rate is as a function of current valuations?

The intuition is that when valuations are high, expected future return is lower, so the maximum safe withdrawal rate should also be lower.

I don't think valuations can inform optimal asset allocation, but I do think they ought to inform optimal retirement point. If the traditional strategy supported by historical data is that one can retire as soon as you have 25x living expenses (i.e. 4% SWR) in savings, then this target may instead be a function of current valuations. e.g. maybe at current market valuations, one should wait until they have 30x, 35x, or possibly 40x living expenses in savings. I don't know. The purpose of this question is to find out.

If no one has conducted this analysis before, I can be the first to contribute an original analysis. Otherwise if someone else has already done it, I would rather not duplicate work.
Michael Kitces did extensive research into this precise topic back in 2008. https://www.kitces.com/wp-content/uploa ... y-2008.pdf

He found that when valuations (CAPE/PE10) were in the highest quintile, the optimal portfolio had an AA of 60/40, and the safe withdrawal rate was 4.4%.

It should be noted that we are at the upper end of the top quintile of historic valuations, which might lower the safe withdrawal rate even more, but we will only know this in hindsight. It should be noted, however, that valuations were much higher in the year 2000, but those who retired that year using the '4% rule' are in good shape right now.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

randomguy
Posts: 5564
Joined: Wed Sep 17, 2014 9:00 am

Re: Valuation-dependent Safe Withdrawal Rate?

Post by randomguy » Mon Feb 12, 2018 2:37 pm

getrichslowly wrote:
Mon Feb 12, 2018 11:52 am

The intuition is that when valuations are high, expected future return is lower, so the maximum safe withdrawal rate should also be lower.
The 4% rule has already accounted for high valuations so there is no need to lower it to account for high valuations. If you can raise your withdrawal rate in normal/low valuation periods would be a more interesting question. But I can't imagine many people would have the stomach for starting with a 6% SWR cause valuations are low:)

getrichslowly
Posts: 73
Joined: Mon Feb 12, 2018 11:48 am

Re: Valuation-dependent Safe Withdrawal Rate?

Post by getrichslowly » Mon Feb 12, 2018 2:40 pm


I think I disagree with this premise. It is basically trying to time the market by changing your asset allocation based on stock valuations. The problem with this strategy is that it is backtested over historical data, using knowledge of historical data to generate the strategy itself. It's a tautology. It relies on the historical artifact that stocks mean revert. The model will fail completely in the future if stocks don't mean revert, or the mean reversion point has shifted due to shifts in the structural parameters.

The focus of my study is to not fiddle with my asset allocation, which I believe should be static, but to have realistic and sober expectations about maximum SWR moving forward.

getrichslowly
Posts: 73
Joined: Mon Feb 12, 2018 11:48 am

Re: Valuation-dependent Safe Withdrawal Rate?

Post by getrichslowly » Mon Feb 12, 2018 2:42 pm

willthrill81 wrote:
Mon Feb 12, 2018 2:20 pm
It should be noted, however, that valuations were much higher in the year 2000, but those who retired that year using the '4% rule' are in good shape right now.
I think that "good shape" is overestimated by the unusual runup in valuations recently. The problem with speculative price increases is that they aren't sustainable forever, yet they trick people into extrapolating past returns into future returns... Put another way, if you retired at the top of the bubble, and the market runs up into a second even greater bigger bubble, then of course on paper it looks okay. But that's really just luck, and it won't last forever, and its not sustainable.

User avatar
HomerJ
Posts: 11058
Joined: Fri Jun 06, 2008 12:50 pm

Re: Valuation-dependent Safe Withdrawal Rate?

Post by HomerJ » Mon Feb 12, 2018 2:45 pm

getrichslowly wrote:
Mon Feb 12, 2018 11:52 am
If the traditional strategy supported by historical data is that one can retire as soon as you have 25x living expenses (i.e. 4% SWR) in savings, then this target may instead be a function of current valuations. e.g. maybe at current market valuations, one should wait until they have 30x, 35x, or possibly 40x living expenses in savings. I don't know. The purpose of this question is to find out.
The "4% SWR" is not an average of all historical SWRs.

4% worked in the worst times.

Looking at history, we see something like this

20% of the time - you could withdraw 7-8% and not run out of money in 30 years
60% of the time - you could withdraw 5-6% and not run out of money in 30 years
20% of the time - you could withdraw 4% and not run out of money in 30 years

So people looked at that data, and said "We better go with 4%, because you never know if you are retiring in one of the bottom 20% years."

Now, you're asking if valuations can help give us a prediction if we're retiring into one of the bottom 20% years.

Maybe. It seems reasonable.

But if you're already using 4%, then you already prepared for the worst years. Valuations may tell you that we have a 40% bigger chance to be in a worst year, but you're already ready for that.

I certainly wouldn't use valuations in the OTHER direction. If valuations were low, and maybe that told me we have a 40% bigger chance to be in one of the best years, I certainly wouldn't start taking out 8% a year my first year in retirement. Instead I'd probably still stick with 4%, and adjust to what actually happens instead of what is predicted to happen.

getrichslowly
Posts: 73
Joined: Mon Feb 12, 2018 11:48 am

Re: Valuation-dependent Safe Withdrawal Rate?

Post by getrichslowly » Mon Feb 12, 2018 2:46 pm

randomguy wrote:
Mon Feb 12, 2018 2:37 pm
getrichslowly wrote:
Mon Feb 12, 2018 11:52 am

The intuition is that when valuations are high, expected future return is lower, so the maximum safe withdrawal rate should also be lower.
The 4% rule has already accounted for high valuations so there is no need to lower it to account for high valuations. If you can raise your withdrawal rate in normal/low valuation periods would be a more interesting question. But I can't imagine many people would have the stomach for starting with a 6% SWR cause valuations are low:)
The 4% rule was established over high equation valuations but not high bond valuations. We currently have both, which is a historical abnormality. Also, even the 4% rule has a 5% chance of a failure baked in, so it was never perfect. So if you are retiring on the top of end valuations, you can probably expect an even greater than 5% chance of failure, based on basic statistics. Put another way, while the unconditional probability of failure is 5%, the conditional probability, i.e conditional on valuations, will be a function of valuations. If valuations are higher than average, the probability of failure may be greater than 5%. If the valuations are lower than average, the probability of failure may be less 5%. (All assuming a constant 4% SWR). Thus if you want to restore a 5% chance of failure when valuations are high, you need to reduce your SWR to reclaim 95% odds of success.

User avatar
willthrill81
Posts: 4063
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Valuation-dependent Safe Withdrawal Rate?

Post by willthrill81 » Mon Feb 12, 2018 2:49 pm

getrichslowly wrote:
Mon Feb 12, 2018 2:42 pm
willthrill81 wrote:
Mon Feb 12, 2018 2:20 pm
It should be noted, however, that valuations were much higher in the year 2000, but those who retired that year using the '4% rule' are in good shape right now.
I think that "good shape" is overestimated by the unusual runup in valuations recently. The problem with speculative price increases is that they aren't sustainable forever, yet they trick people into extrapolating past returns into future returns... Put another way, if you retired at the top of the bubble, and the market runs up into a second even greater bigger bubble, then of course on paper it looks okay. But that's really just luck, and it won't last forever, and its not sustainable.
You're losing sight of the inherent point of safe withdrawal rate research: to find what is the maximum rate that would have been sustained for a certain period of time in the past. No one knows whether the future will look the same as the past, and that's a common criticism of this entire line of research. But we have to work with what we have without the benefit of crystal balls.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

User avatar
HomerJ
Posts: 11058
Joined: Fri Jun 06, 2008 12:50 pm

Re: Valuation-dependent Safe Withdrawal Rate?

Post by HomerJ » Mon Feb 12, 2018 3:07 pm

getrichslowly wrote:
Mon Feb 12, 2018 2:46 pm
The 4% rule was established over high equation valuations but not high bond valuations. We currently have both, which is a historical abnormality.
Where did you get that idea?

Bond rates were lower than today in 1946, and stock market valuations were high. Yet 4% worked for people retiring in 1946.

Bond rates were higher than today in 1966, but interest rates increased from 1966 to 1982, AND we had high inflation from 1977-1982, AND the stock market went nowhere during the same period, and still 4% worked.

Okay, technically it was 3.8% I believe that worked if you retired in 1966.

But 1966 was the absolute worst year so far. And BOTH bonds AND stocks did terrible for the first 16 years of your retirement. AND we had 5 years of high inflation in there too.

And still 4% (almost) worked.

So 4% is pretty conservative.

Besides, it's not an iron-clad RULE... It's a guide. If the market crashes 2 years into your retirement, you don't have to keep pulling the full 4% out every year. You can cut back on trips or wait to buy a new car, etc. You can adjust.

As long as 4% doesn't represent bare-bones survival, I would think that 4% is plenty conservative.

But if you want to go down to 3.5% or 3% for even more safety... that's fine... There are actually quite a few people here who agree with you.

But I personally think 4% (with some of those expenses being discretionary) is plenty conservative.

getrichslowly
Posts: 73
Joined: Mon Feb 12, 2018 11:48 am

Re: Valuation-dependent Safe Withdrawal Rate?

Post by getrichslowly » Mon Feb 12, 2018 3:31 pm

HomerJ wrote:
Mon Feb 12, 2018 3:07 pm
getrichslowly wrote:
Mon Feb 12, 2018 2:46 pm
The 4% rule was established over high equation valuations but not high bond valuations. We currently have both, which is a historical abnormality.
Where did you get that idea?

Bond rates were lower than today in 1946, and stock market valuations were high. Yet 4% worked for people retiring in 1946.
CAPE in 1946 was 16. Today it is 32. It is double.

So maybe I will start the analysis. I will probably do a regression on two variables: bond valuations and stock valuations, and see what the maximum SWR is. From that maybe I can predict what the maximum SWR would be using today's valuations.

While we have historical data with high stock valuations and high bond valuations, we haven't had historical data with both at the same time. Today is an unusual moment in history.

User avatar
HomerJ
Posts: 11058
Joined: Fri Jun 06, 2008 12:50 pm

Re: Valuation-dependent Safe Withdrawal Rate?

Post by HomerJ » Mon Feb 12, 2018 3:57 pm

getrichslowly wrote:
Mon Feb 12, 2018 3:31 pm
HomerJ wrote:
Mon Feb 12, 2018 3:07 pm
getrichslowly wrote:
Mon Feb 12, 2018 2:46 pm
The 4% rule was established over high equation valuations but not high bond valuations. We currently have both, which is a historical abnormality.
Where did you get that idea?

Bond rates were lower than today in 1946, and stock market valuations were high. Yet 4% worked for people retiring in 1946.
CAPE in 1946 was 16. Today it is 32. It is double.

So maybe I will start the analysis. I will probably do a regression on two variables: bond valuations and stock valuations, and see what the maximum SWR is. From that maybe I can predict what the maximum SWR would be using today's valuations.

While we have historical data with high stock valuations and high bond valuations, we haven't had historical data with both at the same time. Today is an unusual moment in history.
You are correct about CAPE... I was looking at PE, not PE10. PE ratio in 1946 was 19, which was quite high at the time, and bond rates were super low.

I'm confused when you say "high bond valuations". What exactly does that mean? Interest rates are low right now, yes... but I've never heard the expression "bond valuations".

And again, today may be an unusual moment in history, but 4% worked during the Great Depression, and it worked during 1966-1982, which was a REALLY bad 16-year stretch for both bonds and stocks, with double-digit inflation at the end.

4% is a pretty conservative SWR.

If getting to 33x expenses (3% SWR) doesn't cause you too much sacrifice, and helps you sleep at night, it sounds good to me.

If you're 63, and you have to work another 5 years to get from 25x to 33x, remember that running out of time is also a risk, just like running out of money is a risk.

Me, I'd suggest blending the two SWRs. Use a 4% withdrawal, but have the ability and willingness to drop back to 3% if things aren't going well.

SPIAs also become more attractive, if you're wanting to use a 3% SWR.

randomguy
Posts: 5564
Joined: Wed Sep 17, 2014 9:00 am

Re: Valuation-dependent Safe Withdrawal Rate?

Post by randomguy » Mon Feb 12, 2018 8:01 pm

getrichslowly wrote:
Mon Feb 12, 2018 3:31 pm
HomerJ wrote:
Mon Feb 12, 2018 3:07 pm
getrichslowly wrote:
Mon Feb 12, 2018 2:46 pm
The 4% rule was established over high equation valuations but not high bond valuations. We currently have both, which is a historical abnormality.
Where did you get that idea?

Bond rates were lower than today in 1946, and stock market valuations were high. Yet 4% worked for people retiring in 1946.
CAPE in 1946 was 16. Today it is 32. It is double.

So maybe I will start the analysis. I will probably do a regression on two variables: bond valuations and stock valuations, and see what the maximum SWR is. From that maybe I can predict what the maximum SWR would be using today's valuations.

While we have historical data with high stock valuations and high bond valuations, we haven't had historical data with both at the same time. Today is an unusual moment in history.
But today’s cape10 also isn’t 1946s CAPE10 either as the definition on earnings has changed. 1937 had a 21 PE10 and 2.6 interest rates. That’s in the ballpark. Remember back then the cape10 average was a lot lower (just over 10 springs to mind but I can’t remuneration if that was from 1906 or 1880). The real question is obviously what will happen to CAPE10 going forward. We are like 30+ years of being above average. Either something’s changed or we are getting some huge correction in the future.

Given we have no data, how are you going to make predications? It isn’t like you can treat the bond market and stock market as independent variables given the interactions between the two. And we haven’t even talked about predicting inflation which is what made the 70s extra special.

AlohaJoe
Posts: 3206
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Valuation-dependent Safe Withdrawal Rate?

Post by AlohaJoe » Mon Feb 12, 2018 8:31 pm

getrichslowly wrote:
Mon Feb 12, 2018 2:46 pm
The 4% rule was established over high equation valuations but not high bond valuations. We currently have both, which is a historical abnormality. Also, even the 4% rule has a 5% chance of a failure baked in, so it was never perfect. So if you are retiring on the top of end valuations, you can probably expect an even greater than 5% chance of failure, based on basic statistics. Put another way, while the unconditional probability of failure is 5%, the conditional probability, i.e conditional on valuations, will be a function of valuations.
This is all true enough but it is also hard to know how much confidence we should have in a belief that our future will be uniquely bad.

People have tried to do what you suggest. In 2013 three well-known retirement researchers build a model of future returns that accounts for both current (especially at the time) low-yields and high valuations. The resulting paper, "Deconstructing Safe Withdrawal Rates in a Low-Yield World" had a headline result that a 60/40 portfolio may only have a 3.2% safe withdrawal rate given current yields & valuations.

The "problem" is that people have been crying about low-bond yields ruining the world and high valuations ruining the world for, I dunno, eight or nine years now?

So far not a single one of their predictions has come true about a single thing. (Remember when the same people were telling us about inflation or even hyperinflation?)

The question you are asking is a hard one. You are asking people to build a model of how the current yield & valuation situation will progress over the next 20-30 years. But so far, all of the "standard models" have gotten it totally wrong over the past 10 years. What if valuations never regress? What if yields never regress? What volatility stays unnaturally low for two decades? (Volatility is the single biggest impact on safe withdrawal rates, so with low volatility it doesn't matter if yields are low or equity returns are low.)

User avatar
zTurtle
Posts: 6
Joined: Mon Feb 12, 2018 12:38 pm

Re: Valuation-dependent Safe Withdrawal Rate?

Post by zTurtle » Mon Feb 12, 2018 8:32 pm

HomerJ wrote:
Mon Feb 12, 2018 2:45 pm
The "4% SWR" is not an average of all historical SWRs.
4% worked in the worst times.
Looking at history, we see something like this
20% of the time - you could withdraw 7-8% and not run out of money in 30 years
60% of the time - you could withdraw 5-6% and not run out of money in 30 years
20% of the time - you could withdraw 4% and not run out of money in 30 years
So people looked at that data, and said "We better go with 4%, because you never know if you are retiring in one of the bottom 20% years."
This is a great summary thank you!

OP, if I follow you're trying to prove a correlation between (a) failure of the 4% SWR historically, during all the 30-year periods that we have data, and (b) high valuation of both stocks and bonds in the first year of the period. If the correlation holds, you can then use this formula to get a probability of failure for the current stock/bond valuations. Seems reasonable, except I don't think these are independent variables. The stock/bond valuations in the first year of the period is know by the market and will influence investor behavior, so I'm not sure you can separate it out and then use it as a predictor. But I'll admit my understanding of regression analysis is weak.

Also there is the bigger issue that you're essentially trying to come up with a formula for predicting whether the current year is a "good" time to invest (where "good" is defined as SWR=4% not failing in the next 30 years) which sounds like "marking timing" to me.

getrichslowly
Posts: 73
Joined: Mon Feb 12, 2018 11:48 am

Re: Valuation-dependent Safe Withdrawal Rate?

Post by getrichslowly » Tue Feb 13, 2018 10:01 am

randomguy wrote:
Mon Feb 12, 2018 8:01 pm
getrichslowly wrote:
Mon Feb 12, 2018 3:31 pm
HomerJ wrote:
Mon Feb 12, 2018 3:07 pm
getrichslowly wrote:
Mon Feb 12, 2018 2:46 pm
The 4% rule was established over high equation valuations but not high bond valuations. We currently have both, which is a historical abnormality.
Where did you get that idea?

Bond rates were lower than today in 1946, and stock market valuations were high. Yet 4% worked for people retiring in 1946.
CAPE in 1946 was 16. Today it is 32. It is double.

So maybe I will start the analysis. I will probably do a regression on two variables: bond valuations and stock valuations, and see what the maximum SWR is. From that maybe I can predict what the maximum SWR would be using today's valuations.

While we have historical data with high stock valuations and high bond valuations, we haven't had historical data with both at the same time. Today is an unusual moment in history.
But today’s cape10 also isn’t 1946s CAPE10 either as the definition on earnings has changed. 1937 had a 21 PE10 and 2.6 interest rates. That’s in the ballpark. Remember back then the cape10 average was a lot lower (just over 10 springs to mind but I can’t remuneration if that was from 1906 or 1880). The real question is obviously what will happen to CAPE10 going forward. We are like 30+ years of being above average. Either something’s changed or we are getting some huge correction in the future.

Given we have no data, how are you going to make predications? It isn’t like you can treat the bond market and stock market as independent variables given the interactions between the two. And we haven’t even talked about predicting inflation which is what made the 70s extra special.
Bond markets are actually simpler than stocks as the E[R] is simply equal to the current interest yield. Whereas with stocks we have a loose idea that higher valuations lower future earnings, but technically there is nothing stopping earnings from growing, and the high valuation actually being justified.

getrichslowly
Posts: 73
Joined: Mon Feb 12, 2018 11:48 am

Re: Valuation-dependent Safe Withdrawal Rate?

Post by getrichslowly » Tue Feb 13, 2018 10:02 am

zTurtle wrote:
Mon Feb 12, 2018 8:32 pm
HomerJ wrote:
Mon Feb 12, 2018 2:45 pm
The "4% SWR" is not an average of all historical SWRs.
4% worked in the worst times.
Looking at history, we see something like this
20% of the time - you could withdraw 7-8% and not run out of money in 30 years
60% of the time - you could withdraw 5-6% and not run out of money in 30 years
20% of the time - you could withdraw 4% and not run out of money in 30 years
So people looked at that data, and said "We better go with 4%, because you never know if you are retiring in one of the bottom 20% years."
This is a great summary thank you!

OP, if I follow you're trying to prove a correlation between (a) failure of the 4% SWR historically, during all the 30-year periods that we have data, and (b) high valuation of both stocks and bonds in the first year of the period. If the correlation holds, you can then use this formula to get a probability of failure for the current stock/bond valuations. Seems reasonable, except I don't think these are independent variables. The stock/bond valuations in the first year of the period is know by the market and will influence investor behavior, so I'm not sure you can separate it out and then use it as a predictor. But I'll admit my understanding of regression analysis is weak.

Also there is the bigger issue that you're essentially trying to come up with a formula for predicting whether the current year is a "good" time to invest (where "good" is defined as SWR=4% not failing in the next 30 years) which sounds like "marking timing" to me.
I don't time the market. I only want to have an unbiased estimate of E[R] so I can adjust my withdrawals downward if needed.

getrichslowly
Posts: 73
Joined: Mon Feb 12, 2018 11:48 am

Re: Valuation-dependent Safe Withdrawal Rate?

Post by getrichslowly » Tue Feb 13, 2018 10:03 am

randomguy wrote:
Mon Feb 12, 2018 8:01 pm
getrichslowly wrote:
Mon Feb 12, 2018 3:31 pm
HomerJ wrote:
Mon Feb 12, 2018 3:07 pm
getrichslowly wrote:
Mon Feb 12, 2018 2:46 pm
The 4% rule was established over high equation valuations but not high bond valuations. We currently have both, which is a historical abnormality.
Where did you get that idea?

Bond rates were lower than today in 1946, and stock market valuations were high. Yet 4% worked for people retiring in 1946.
CAPE in 1946 was 16. Today it is 32. It is double.

So maybe I will start the analysis. I will probably do a regression on two variables: bond valuations and stock valuations, and see what the maximum SWR is. From that maybe I can predict what the maximum SWR would be using today's valuations.

While we have historical data with high stock valuations and high bond valuations, we haven't had historical data with both at the same time. Today is an unusual moment in history.
But today’s cape10 also isn’t 1946s CAPE10 either as the definition on earnings has changed. 1937 had a 21 PE10 and 2.6 interest rates. That’s in the ballpark. Remember back then the cape10 average was a lot lower (just over 10 springs to mind but I can’t remuneration if that was from 1906 or 1880). The real question is obviously what will happen to CAPE10 going forward. We are like 30+ years of being above average. Either something’s changed or we are getting some huge correction in the future.

Given we have no data, how are you going to make predications? It isn’t like you can treat the bond market and stock market as independent variables given the interactions between the two. And we haven’t even talked about predicting inflation which is what made the 70s extra special.
Can you elaborate or point to sources on how the definition on earnings has changed? Not that I doubt you, as I'm familiar with data reporting standards changing over time for other variables like income inequality, labor:capital income share ratio, etc., but I'm curious to see this in more detail, as this could completely revolutionize the CAPE historical statistics that everyone cites.

User avatar
HomerJ
Posts: 11058
Joined: Fri Jun 06, 2008 12:50 pm

Re: Valuation-dependent Safe Withdrawal Rate?

Post by HomerJ » Tue Feb 13, 2018 10:09 am

getrichslowly wrote:
Tue Feb 13, 2018 10:03 am
Can you elaborate or point to sources on how the definition on earnings has changed? Not that I doubt you, as I'm familiar with data reporting standards changing over time for other variables like income inequality, labor:capital income share ratio, etc., but I'm curious to see this in more detail, as this could completely revolutionize the CAPE historical statistics that everyone cites.
Here's an absolutely excellent article about CAPE and the accounting changes.

http://www.philosophicaleconomics.com/2013/12/shiller/

User avatar
jeffyscott
Posts: 6785
Joined: Tue Feb 27, 2007 9:12 am
Location: Wisconsin

Re: Valuation-dependent Safe Withdrawal Rate?

Post by jeffyscott » Tue Feb 13, 2018 11:09 am

HomerJ wrote:
Mon Feb 12, 2018 2:45 pm
Looking at history, we see something like this

20% of the time - you could withdraw 7-8% and not run out of money in 30 years
60% of the time - you could withdraw 5-6% and not run out of money in 30 years
20% of the time - you could withdraw 4% and not run out of money in 30 years

So people looked at that data, and said "We better go with 4%, because you never know if you are retiring in one of the bottom 20% years."
So if at "normal" valuations one might say 5-6% would be okay. Lets assume a 50% decline in stocks and maybe 10% in bonds gets us to normal. If you have a 50/50 portfolio, you would lose 30% in that scenario. If after a 30% loss, 5% would be "safe", this means 3.5% should be safe now. If 6% would be safe when starting from normal valuations, then 4.2% would be safe now.
press on, regardless - John C. Bogle

AlohaJoe
Posts: 3206
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Valuation-dependent Safe Withdrawal Rate?

Post by AlohaJoe » Tue Feb 13, 2018 11:16 am

jeffyscott wrote:
Tue Feb 13, 2018 11:09 am
So if at "normal" valuations one might say 5-6% would be okay. Lets assume a 50% decline in stocks and maybe 10% in bonds gets us to normal. If you have a 50/50 portfolio, you would lose 30% in that scenario. If after a 30% loss, 5% would be "safe", this means 3.5% should be safe now. If 6% would be safe when starting from normal valuations, then 4.2% would be safe now.
None of that makes any sense to me. That isn't how safe withdrawal rates are calculated.

getrichslowly
Posts: 73
Joined: Mon Feb 12, 2018 11:48 am

Re: Valuation-dependent Safe Withdrawal Rate?

Post by getrichslowly » Tue Feb 13, 2018 2:22 pm

AlohaJoe wrote:
Tue Feb 13, 2018 11:16 am
jeffyscott wrote:
Tue Feb 13, 2018 11:09 am
So if at "normal" valuations one might say 5-6% would be okay. Lets assume a 50% decline in stocks and maybe 10% in bonds gets us to normal. If you have a 50/50 portfolio, you would lose 30% in that scenario. If after a 30% loss, 5% would be "safe", this means 3.5% should be safe now. If 6% would be safe when starting from normal valuations, then 4.2% would be safe now.
None of that makes any sense to me. That isn't how safe withdrawal rates are calculated.
The intuition is that if you assume the market will correct itself to its historical mean valuation, you can calculate the fundamental value of your holdings, and base your SWR off that. This way you tune out speculative noise and focus on fundamentals.

It doesn't work the other way unfortunately -- if the market is undervalued, you can't inflate your SWR, because you'd be liquidating capital that needs to grow during a future correction upwards.

User avatar
willthrill81
Posts: 4063
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Valuation-dependent Safe Withdrawal Rate?

Post by willthrill81 » Tue Feb 13, 2018 2:47 pm

getrichslowly wrote:
Tue Feb 13, 2018 2:22 pm
AlohaJoe wrote:
Tue Feb 13, 2018 11:16 am
jeffyscott wrote:
Tue Feb 13, 2018 11:09 am
So if at "normal" valuations one might say 5-6% would be okay. Lets assume a 50% decline in stocks and maybe 10% in bonds gets us to normal. If you have a 50/50 portfolio, you would lose 30% in that scenario. If after a 30% loss, 5% would be "safe", this means 3.5% should be safe now. If 6% would be safe when starting from normal valuations, then 4.2% would be safe now.
None of that makes any sense to me. That isn't how safe withdrawal rates are calculated.
The intuition is that if you assume the market will correct itself to its historical mean valuation, you can calculate the fundamental value of your holdings, and base your SWR off that. This way you tune out speculative noise and focus on fundamentals.

It doesn't work the other way unfortunately -- if the market is undervalued, you can't inflate your SWR, because you'd be liquidating capital that needs to grow during a future correction upwards.
Actually, Kitces found that the opposite was true: when valuations were low, the subsequent safe withdrawal rate was substantially higher than otherwise, well over 5%.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

User avatar
jeffyscott
Posts: 6785
Joined: Tue Feb 27, 2007 9:12 am
Location: Wisconsin

Re: Valuation-dependent Safe Withdrawal Rate?

Post by jeffyscott » Tue Feb 13, 2018 3:02 pm

getrichslowly wrote:
Tue Feb 13, 2018 2:22 pm
AlohaJoe wrote:
Tue Feb 13, 2018 11:16 am
jeffyscott wrote:
Tue Feb 13, 2018 11:09 am
So if at "normal" valuations one might say 5-6% would be okay. Lets assume a 50% decline in stocks and maybe 10% in bonds gets us to normal. If you have a 50/50 portfolio, you would lose 30% in that scenario. If after a 30% loss, 5% would be "safe", this means 3.5% should be safe now. If 6% would be safe when starting from normal valuations, then 4.2% would be safe now.
None of that makes any sense to me. That isn't how safe withdrawal rates are calculated.
The intuition is that if you assume the market will correct itself to its historical mean valuation, you can calculate the fundamental value of your holdings, and base your SWR off that. This way you tune out speculative noise and focus on fundamentals.

It doesn't work the other way unfortunately -- if the market is undervalued, you can't inflate your SWR, because you'd be liquidating capital that needs to grow during a future correction upwards.
Maybe the math would help explain the idea. If you were to lose 30%, $100 would become $70. Taking 5-6% of $70 would be $3.50 to $4.20, which is 3.5-4.2% of the current value of $100.

Seems like the undervalued situation is something we may never need to worry much about, if we take normal as CAPE of about 16. Haven't been below that in around 30 years, aside from about 3 seconds in 2009.
press on, regardless - John C. Bogle

User avatar
willthrill81
Posts: 4063
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Valuation-dependent Safe Withdrawal Rate?

Post by willthrill81 » Tue Feb 13, 2018 3:14 pm

jeffyscott wrote:
Tue Feb 13, 2018 3:02 pm
getrichslowly wrote:
Tue Feb 13, 2018 2:22 pm
AlohaJoe wrote:
Tue Feb 13, 2018 11:16 am
jeffyscott wrote:
Tue Feb 13, 2018 11:09 am
So if at "normal" valuations one might say 5-6% would be okay. Lets assume a 50% decline in stocks and maybe 10% in bonds gets us to normal. If you have a 50/50 portfolio, you would lose 30% in that scenario. If after a 30% loss, 5% would be "safe", this means 3.5% should be safe now. If 6% would be safe when starting from normal valuations, then 4.2% would be safe now.
None of that makes any sense to me. That isn't how safe withdrawal rates are calculated.
The intuition is that if you assume the market will correct itself to its historical mean valuation, you can calculate the fundamental value of your holdings, and base your SWR off that. This way you tune out speculative noise and focus on fundamentals.

It doesn't work the other way unfortunately -- if the market is undervalued, you can't inflate your SWR, because you'd be liquidating capital that needs to grow during a future correction upwards.
Maybe the math would help explain the idea. If you were to lose 30%, $100 would become $70. Taking 5-6% of $70 would be $3.50 to $4.20, which is 3.5-4.2% of the current value of $100.

Seems like the undervalued situation is something we may never need to worry much about, if we take normal as CAPE of about 16. Haven't been below that in around 30 years, aside from about 3 seconds in 2009.
What you are ignoring is the rather strong mean reversion tendency of stocks. When stocks have just dropped in value by 30%, they have a higher expected return going forward than when stocks have recently soared by 30%.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

randomguy
Posts: 5564
Joined: Wed Sep 17, 2014 9:00 am

Re: Valuation-dependent Safe Withdrawal Rate?

Post by randomguy » Tue Feb 13, 2018 6:16 pm

getrichslowly wrote:
Tue Feb 13, 2018 10:01 am
Bond markets are actually simpler than stocks as the E[R] is simply equal to the current interest yield. Whereas with stocks we have a loose idea that higher valuations lower future earnings, but technically there is nothing stopping earnings from growing, and the high valuation actually being justified.
And how well did using that rule work at predicting the real return of bonds in the 1966-1981 time frame (huge inflation) or 1929-1939 (deflation)? SWR are driven by real returns not nominal ones.

JBTX
Posts: 3112
Joined: Wed Jul 26, 2017 12:46 pm

Re: Valuation-dependent Safe Withdrawal Rate?

Post by JBTX » Tue Feb 13, 2018 6:22 pm

getrichslowly wrote:
Mon Feb 12, 2018 2:40 pm

I think I disagree with this premise. It is basically trying to time the market by changing your asset allocation based on stock valuations.The problem with this strategy is that it is backtested over historical data, using knowledge of historical data to generate the strategy itself. It's a tautology. It relies on the historical artifact that stocks mean revert. The model will fail completely in the future if stocks don't mean revert, or the mean reversion point has shifted due to shifts in the structural parameters.

The focus of my study is to not fiddle with my asset allocation, which I believe should be static, but to have realistic and sober expectations about maximum SWR moving forward.

You mean like concluding a 4% withdrawal rate is “safe” based solely on US historical data?

User avatar
jeffyscott
Posts: 6785
Joined: Tue Feb 27, 2007 9:12 am
Location: Wisconsin

Re: Valuation-dependent Safe Withdrawal Rate?

Post by jeffyscott » Tue Feb 13, 2018 6:25 pm

willthrill81 wrote:
Tue Feb 13, 2018 3:14 pm
What you are ignoring is the rather strong mean reversion tendency of stocks. When stocks have just dropped in value by 30%, they have a higher expected return going forward than when stocks have recently soared by 30%.
I don't think so. Looking at it in reverse, I'm suggesting a 3.5-4.2% withdrawal rate right now becoming a 5-6% withdrawal rate after a 50% decline in stock prices and 10% in bond prices for total decline of 30% in a 50/50 portfolio.
press on, regardless - John C. Bogle

User avatar
willthrill81
Posts: 4063
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Valuation-dependent Safe Withdrawal Rate?

Post by willthrill81 » Tue Feb 13, 2018 6:31 pm

jeffyscott wrote:
Tue Feb 13, 2018 6:25 pm
willthrill81 wrote:
Tue Feb 13, 2018 3:14 pm
What you are ignoring is the rather strong mean reversion tendency of stocks. When stocks have just dropped in value by 30%, they have a higher expected return going forward than when stocks have recently soared by 30%.
I don't think so. Looking at it in reverse, I'm suggesting a 3.5-4.2% withdrawal rate right now becoming a 5-6% withdrawal rate after a 50% decline in stock prices and 10% in bond prices for total decline of 30% in a 50/50 portfolio.
If that were a real concern, then a 4% withdrawal rate would have failed many times in the historic record, every time that retirees experienced a large stock decline early in their retirement.

Mean reversion is a very real and well documented phenomenon. Stocks that are deeply 'discounted' compared to normal valuations have a propensity to have higher returns going forward.

Kitces reported that low valuations have historically led to significantly higher SWRs back in 2008.
A straightforward set of starting enhancements to the current safe withdrawal rate rules begins to emerge from the data in Figures 7 and 8. Although there are a wide range of results for markets with P/E10 ratios in the middle quintiles, at a minimum it appears that clients can safely add 0.5% to their safe withdrawal rate as long as the market’s P/E10 is NOT in the worst quintile. In other words, the only instances in history that a safe withdrawal rate below 4.5% was necessary all occurred in environments that had unusually high starting P/E10 valuations. As long as this is not the valuation situation for the prospective retiree, a higher safe withdrawal rate appears to be reliably sustainable.

In addition, the data also reveal that the reverse situation is relevant as well – that when markets are at extremely undervalued levels as measured by P/E10, a significantly higher safe withdrawal rate is merited. Specifically, in all historical market situations where the starting P/E10 was below 12.0, a withdrawal rate upwards of 5.5% (actually, 5.7% - 5.8% by this data) was safe in all historical scenarios as long as equity exposure was at least 60%.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

Post Reply