# The Money Mustache Community

## Learning, Sharing, and Teaching => Investor Alley => Topic started by: SearchingForSimple on March 20, 2015, 07:56:32 PM

Title: Using market valuation measurements to affect SWR?
Post by: SearchingForSimple on March 20, 2015, 07:56:32 PM
I ran some analysis using cfiresim.com coupled with the CAPE data available at multpl.com. Using the standard cfiresim starting point (\$1M portfolio, \$40K expenditures, 30 year timeline and 0.18% Exp Ratio) I changed the allocation to 80/20 stocks and ran the simulation. Only 8 starting years failed out of all those 30 year cycles: 1906, 1964-69, and 1973. 93% success rate - not too shabby.

Looking at the CAPE data for those failed years, we get the following results (as of Jan 1st of each year):
1906 - 20.10
1964 - 21.63
1965 - 23.27
1966 - 24.06
1967 - 20.43
1968 - 21.51
1969 - 21.19
1973 - 18.71 - lowest CAPE in failure data set

Given that the historical CAPE mean is 16.59, these years are all above that average, which isn't too surprising. What was surprising to me though is that there are only 12 other years which have a CAPE above 18.7 in the whole data set (1881 through 1985 since 1985 is the "last" 30 year cycle starting point we have available). That raw math suggests that a 4% SWR rate will fail 40% of the time (8 failed years / 20 total years) when started in a year when the CAPE is above 18.7.

Now there's probably some logical and statistical issues in that analysis, but I point it out because I don't know that I would feel comfortable jumping into retirement using the 4% SWR assumption if there are indications that the market may be overvalued.

Does anyone else worry about this? Anybody planning on altering their SWR based on potential market valuation concerns?
Title: Re: Using market valuation measurements to affect SWR?
Post by: forummm on March 20, 2015, 08:08:48 PM
Yes, I would not retire right now and expect that a 4% WR would be as bulletproof as other times in history. Depending on your metric, the US market is anywhere from 20% to 75% overvalued compared to long-term historical returns. Just a decline to historical average would be painful to someone who retired now. A decline to, say, 25% below average (which has happened for whole decades) would be devastating for someone who retires now and expects a 4% spend rate for 50 years.

I don't know the last time (if ever) we had interest rates this low for this long. The US market is the only place that's provided a return the past 6 years. Maybe when rates go back up and bonds start to look more attractive, equity valuations will drop as money flows to those returns.
Title: Re: Using market valuation measurements to affect SWR?
Post by: skyrefuge on March 20, 2015, 08:55:33 PM
Solid first post!

I would only point out that the post-1985 years are looking a bit kinder to high CAPEs.

1993 (CAPE=20.32) has more than 2x the portfolio's starting value, with only 7 years left to the finish line.
1997 (CAPE=28.33) is still above its starting value, 18 years in.

Even the Y2K retiree at peak market insanity (CAPE=43.77) still has about half his starting value, with the race half over. cFIREsim shows a 66% chance of success for the remaining 15 years, so actually better than your 40% failure rate for CAPE>18.71.

So sure, retiring into an overvalued market is certainly a concern, the difficulty is knowing when a market is "overvalued". CAPE worked pretty well pre-1985, but it's less-obvious that it's a valuable metric for the current era.

All that said, at least a part of the reason I'm still working is because of the current market prices.
Title: Re: Using market valuation measurements to affect SWR?
Post by: Indexer on March 20, 2015, 09:20:43 PM
This does concern me some if the goal is a very long retirement.

Most market valuation measures aren't nearly accurate enough to make decisions based on.  CAPE is useless for short time frames(even Shiller will tell you that), but it has shown to be a 'decent' predictor of long term(10yr) stock market returns.  A high CAPE does imply that future returns at least for the next 10 years might not be as high as historic returns.

In addition to the stock portion of a portfolio being at risk of lower future returns bonds have historically had returns in line with their 'yields' which are also at historic lows.  If both stocks and bonds had lower future returns in the that could screw up a 4% SWR.
Title: Re: Using market valuation measurements to affect SWR?
Post by: hodedofome on March 20, 2015, 10:10:15 PM
I'd rather time the SWR to whether or not the market is above or below the 200 day moving average or something objective like that.

You can't trade PE ratios or economic indicators but you can trade price. Therefore timing with price only indicators is IMO the most robust way to do it.
Title: Re: Using market valuation measurements to affect SWR?
Post by: brooklynguy on March 21, 2015, 01:42:15 PM
Great analysis, SearchingForSimple.

It's interesting to note that, using your parameters, 3.4% (really 3.399463%) was the maximum WR that resulted in "success" (i.e., avoidance of total portfolio depletion, but only by a hair) for a person retiring in 1966 (the year with the highest CAPE on record of all the starting years for fully-elapsed 30-year periods).  1966 was the absolute worst time in history to commence a 30-year retirement, so 3.4% is the maximum WR that resulted in a 100% success rate across all historical periods (which can be confirmed by running a "maximum yearly/initial spending for success" investigation analysis in cfiresim).

Increasing the equity allocation above 80% increases that absolute safe-max WR a bit (but only a bit).  Lowering the expense ratio has a bigger effect -- dialing it down from 0.18% to 0.05% (the current VTSAX expense ratio) (and still sticking with the rest of your parameters) improves the absolute safe-max WR to 3.45%.
Title: Re: Using market valuation measurements to affect SWR?
Post by: innerscorecard on March 22, 2015, 08:12:49 PM
There are a lot of problems with simple CAPE. Beware false precision.
Title: Re: Using market valuation measurements to affect SWR?
Post by: sirdoug007 on March 22, 2015, 09:13:01 PM
The problem with this analysis using CAPE > 20 as a threshold is that CAPE has been under 20 only once since 1993, during the 2009 Great Recession.

So other than 2009, for the last 20+ you would have to accept a painfully low withdrawal rate.  I just don't see how that is practical.

Sent from my iPhone using Tapatalk
Title: Re: Using market valuation measurements to affect SWR?
Post by: innerscorecard on March 22, 2015, 09:14:43 PM
Philosophical Economics puts it well:

"We conclude with the question that all of this exists to answer: Is the market expensive? Yes, and returns are likely to be below the historical average, pulled down by a number of different mechanisms.  Should the market be expensive?  “Should” is not an appropriate word to use in markets.  What matters is that there are secular, sustainable forces behind the market’s expensiveness–to name a few: low real interest rates, a lack of alternative investment opportunities (TINA), aggressive policymaker support, and improved market efficiency yielding a reduced equity risk premium (difference between equity returns and fixed income returns).  Unlike in prior eras of history, the secret of “stocks for the long run” is now very well known–thoroughly studied by academics all over the world, and permanently seared into the brain of every investor that sets foot on Wall Street.  For this reason, absent extreme levels of cyclically-induced fear, investors simply aren’t going to foolishly sell equities at bargain prices when there’s nowhere else to go–as they did, for example, in the 1940s and 1950s, when they had limited history and limited studied knowledge on which to rely.

As for the future, the interest-rate-related forces that are pushing up on valuations will get pulled out from under the market if and when inflation ties the Fed’s hands–i.e., forces the Fed to impose a higher real interest rate on the economy.  For all we know, that may never happen.  Similarly, on a cyclically-adjusted basis, the equity risk premium may never again return to what it was in prior periods, as secrets cannot be taken back."
Title: Re: Using market valuation measurements to affect SWR?
Post by: sirdoug007 on April 02, 2015, 02:44:17 PM
I found this chart of 10 year subsequent returns vs. the CAPE in the beginning year from a Wade Pfau paper.  Looking at this scatter I don't see how CAPE is at all useful for FIRE planning purposes.
Title: Re: Using market valuation measurements to affect SWR?
Post by: forummm on April 02, 2015, 04:22:20 PM
I found this chart of 10 year subsequent returns vs. the CAPE in the beginning year from a Wade Pfau paper.  Looking at this scatter I don't see how CAPE is at all useful for FIRE planning purposes.

I have the exact opposite read on this. It looks like there is a clear relationship. Higher CAPE has been pretty clearly associated with lower returns in the next 10 years.
Title: Re: Using market valuation measurements to affect SWR?
Post by: sol on May 06, 2015, 07:41:53 PM
I have the exact opposite read on this. It looks like there is a clear relationship. Higher CAPE has been pretty clearly associated with lower returns in the next 10 years.

The problem with all of these CAPE based metrics is that CAPE hasn't been consistently calculated the same way over time.  The consistent rise above 20 starting in the 90s isn't reflective of everyone suddenly deciding to pay more for stocks, it's due to changes in the way earnings are reported.

The internet is full of chatter on this topic, but here's a sampling for those who really care:

http://www.ft.com/intl/cms/s/0/496a3844-0013-11e3-9c40-00144feab7de.html#axzz3ZPbB14ts
http://www.philosophicaleconomics.com/2013/12/shiller/
http://seekingalpha.com/article/2836966-cape-fear-should-we-brace-for-below-average-returns

With that context in mind, using CAPE to predict future market returns is clearly not going to work in the future the same way that is has in the past.
Title: Re: Using market valuation measurements to affect SWR?
Post by: forummm on May 06, 2015, 07:56:11 PM
I have the exact opposite read on this. It looks like there is a clear relationship. Higher CAPE has been pretty clearly associated with lower returns in the next 10 years.

The problem with all of these CAPE based metrics is that CAPE hasn't been consistently calculated the same way over time.  The consistent rise above 20 starting in the 90s isn't reflective of everyone suddenly deciding to pay more for stocks, it's due to changes in the way earnings are reported.

The internet is full of chatter on this topic, but here's a sampling for those who really care:

http://www.ft.com/intl/cms/s/0/496a3844-0013-11e3-9c40-00144feab7de.html#axzz3ZPbB14ts
http://www.philosophicaleconomics.com/2013/12/shiller/
http://seekingalpha.com/article/2836966-cape-fear-should-we-brace-for-below-average-returns

With that context in mind, using CAPE to predict future market returns is clearly not going to work in the future the same way that is has in the past.

I agree. Philosophicaleconomics has an adjusted CAPE for today that's more like 23-25 for the different ways he calculates it. Still incredibly high, but it shows that the difference is a few points. I think it's still a useful measure if you understand the discrepancy.
Title: Re: Using market valuation measurements to affect SWR?
Post by: brooklynguy on May 07, 2015, 07:20:21 AM
The problem with all of these CAPE based metrics is that CAPE hasn't been consistently calculated the same way over time.  The consistent rise above 20 starting in the 90s isn't reflective of everyone suddenly deciding to pay more for stocks, it's due to changes in the way earnings are reported.

Even putting aside the inconsistencies in the calculation of CAPE over time (which, as the Philosophical Economics author has convincingly argued in his more recent posts, does not matter as much as it might seem to), there are also reasons to believe that there may have been a fundamental shift in the system that can explain why elevated CAPEs have persisted for so long and why they may be sustainable going forward (in other words, why the expected "reversion to the mean" may never arrive).  Arguments that "this time is different" are always dangerous, but there are compelling reasons why they should at least be taken seriously and not dismissed out of hand (but not necessarily accepted, either).

Today's Philosophical Economics post (http://www.philosophicaleconomics.com/2015/05/profit-margins-in-a-winner-take-all-economy/) is an excellent example:  it essentially (and compellingly) argues that the economy has evolved (or is in the process of evolving) into a "winner take all" economy, where barriers to entry prevent the free market's normal competitive forces from reigning in corporate profit margins.
Title: Re: Using market valuation measurements to affect SWR?
Post by: hodedofome on May 07, 2015, 11:24:34 AM
I wouldn't personally use CAPE as a timing device on an individual market, for reasons stated above. But I would possibly use it as a simple measure of value for an individual country relative to other countries. So, instead of using say PE or book to market to find the cheapest countries in the world, I would rather use something like CAPE.

Full disclosure: long a little bit of GVAL in my taxable account
Title: Re: Using market valuation measurements to affect SWR?
Post by: beltim on May 07, 2015, 12:35:31 PM
The problem with all of these CAPE based metrics is that CAPE hasn't been consistently calculated the same way over time.  The consistent rise above 20 starting in the 90s isn't reflective of everyone suddenly deciding to pay more for stocks, it's due to changes in the way earnings are reported.

Even putting aside the inconsistencies in the calculation of CAPE over time (which, as the Philosophical Economics author has convincingly argued in his more recent posts, does not matter as much as it might seem to), there are also reasons to believe that there may have been a fundamental shift in the system that can explain why elevated CAPEs have persisted for so long and why they may be sustainable going forward (in other words, why the expected "reversion to the mean" may never arrive).  Arguments that "this time is different" are always dangerous, but there are compelling reasons why they should at least be taken seriously and not dismissed out of hand (but not necessarily accepted, either).

Today's Philosophical Economics post (http://www.philosophicaleconomics.com/2015/05/profit-margins-in-a-winner-take-all-economy/) is an excellent example:  it essentially (and compellingly) argues that the economy has evolved (or is in the process of evolving) into a "winner take all" economy, where barriers to entry prevent the free market's normal competitive forces from reigning in corporate profit margins.

That seems like an argument for why profit margins may not revert to the mean.  Since CAPE is based on profits, not margins, it doesn't seem like sustained higher profit margins should affect CAPE.

Note, though, that the converse isn't true:  if things weren't different this time, and corporate profit margins were expected to decline to a reversion to the mean, then the CAPE would be unusually/unsustainably high.

At least that's the way I read it.
Title: Re: Using market valuation measurements to affect SWR?
Post by: brooklynguy on May 07, 2015, 01:10:39 PM
That seems like an argument for why profit margins may not revert to the mean.  Since CAPE is based on profits, not margins, it doesn't seem like sustained higher profit margins should affect CAPE.

Note, though, that the converse isn't true:  if things weren't different this time, and corporate profit margins were expected to decline to a reversion to the mean, then the CAPE would be unusually/unsustainably high.

At least that's the way I read it.

Good point.  If there has been a structural change in the system resulting in higher profit margins, CAPE would detect that change and over time incorporate it into itself.  So I guess the author's argument supports the idea of sustained higher profit margins, and therefore sustained higher stock prices, but not higher CAPEs (because CAPE, as a dynamic metric based on earnings/profits, should adapt to the structural change and not incorrectly report the market as being "overvalued").

Right?  I'm starting to confuse myself...
Title: Re: Using market valuation measurements to affect SWR?
Post by: beltim on May 07, 2015, 01:18:36 PM
That seems like an argument for why profit margins may not revert to the mean.  Since CAPE is based on profits, not margins, it doesn't seem like sustained higher profit margins should affect CAPE.

Note, though, that the converse isn't true:  if things weren't different this time, and corporate profit margins were expected to decline to a reversion to the mean, then the CAPE would be unusually/unsustainably high.

At least that's the way I read it.

Good point.  If there has been a structural change in the system resulting in higher profit margins, CAPE would detect that change and over time incorporate it into itself.  So I guess the author's argument supports the idea of sustained higher profit margins, and therefore sustained higher stock prices, but not higher CAPEs (because CAPE, as a dynamic metric based on earnings/profits, should adapt to the structural change and not incorrectly report the market as being "overvalued").

Right?  I'm starting to confuse myself...

Right.  I guess CAPEs could temporarily be high as the higher profit margins cycle in, but that should be completely accounted for 10 years after the high profit margins.  If true, we could be 7-8 into that period.

Title: Re: Using market valuation measurements to affect SWR?
Post by: EscapeVelocity2020 on May 08, 2015, 08:32:29 AM
These have all been fascinating discussions, using a tool like cFIREsim to exact a SWR down to the 3rd decimal point, but I can't help but wonder if the past really is representative of the future (predictive of the next 30 years).  The market didn't have low cost 'index funds' before 1975.  And do we really expect the next world war will happen in our lifetime, and be anything like WWII?  Even in our current state, the Fed has never expanded it's balance sheet and kept rates at roughly zero like this.  So, just maybe, we should take the FIRE simulations off of their high pedestal and spend just as much time talking about what else could be used to judge how prepared we are for 'retirement'.

Unfortunately, there are none that come to mind other than the 'real world experience' on the E-R forum, where most people think we are woefully unprepared.

On the positive side, in my own opinion, it is also a more enjoyable time than any time in history to freelance or be underemployed.  Never before have there been so many inexpensive outlets for collaboration, innovation, or entertainment.  As long as you have confidence and drive, on any given morning you can wake up, think of some novel solution to a problem, and givet a 'kickstarter' campaign a go or document your exciting life on YouTube.   Maybe the market will be flooded on these specific things, but the market for creativity is never flooded.  And if that isn't your thing, there is also the sharing economy where you can now profit off the stuff you already own - rent out part of your house, drive other people around, cook for others, shop for others.  It's lower margin, less X factor, but it buys you new freedom to 'retire' to a different life.

So maybe we are too focused on large 4% SWR number and really only need a 10-year FU money number with the expectation that we find work we truly enjoy and covers the SWR within that 10 years....

Sorry, just Friday free-form ramblings.  You can all go back to playing with simulators and bickering over if a penny of earnings this year is as good a predictor as in was 1965, or whatever :)
Title: Re: Using market valuation measurements to affect SWR?
Post by: EscapeVelocity2020 on May 08, 2015, 08:54:39 AM
By the way, I've always found 'The Great Moderation' to be an interesting theory - http://en.wikipedia.org/wiki/Great_Moderation

(I have to run off to meetings, but I'll check back in after lunch).  Hope my ramblings and postulations get some discussion, if only to say I'm completely off base and 4% SWR is rock solid guaranteed from now until the end of days (and why)...
Title: Re: Using market valuation measurements to affect SWR?
Post by: brooklynguy on May 08, 2015, 10:05:18 AM
These have all been fascinating discussions, using a tool like cFIREsim to exact a SWR down to the 3rd decimal point, but I can't help but wonder if the past really is representative of the future (predictive of the next 30 years) . . . So, just maybe, we should take the FIRE simulations off of their high pedestal and spend just as much time talking about what else could be used to judge how prepared we are for 'retirement'.

Recognition of the false precision inherent in using the past to predict the future, and vacillation between optimism and pessimism about the safety of using historically safe withdrawal rates in the current environment, has kind of been a recurring theme across the forum lately.

In response, I refer to two observations previously made by Skyrefuge and Sol (in my view, two of this forum's greatest thinkers on matters FIRE):

• To paraphrase Skyrefuge (who was in turn paraphrasing Churchill):  "A history-based SWR is the worst form of retirement-readiness predictor, except for all the rest."
• To paraphrase Sol:  "Making imperfect retirement predictions is still better than making no retirement predictions."
Put these two astute insights together, and you have to conclude that there is real value in performing analyses of the type we're doing in this thread (and many others).
Title: Re: Using market valuation measurements to affect SWR?
Post by: EscapeVelocity2020 on May 08, 2015, 10:34:39 AM
I'm not advocating not calculating SWR, but highlighting that it is but a small tool in the big picture.  In fact, without running a single Monte Carlo analysis, I could've said 'I think a million dollars is enough to last me the rest of my life' (which is exactly what I did in 1996 and others before me did - which is kinda funny because you would think inflation would have people increasing that number, but I digress...), since I spend around 40k each year.  Then, just using common sense, I can say 'gosh, two million dollars is pretty much bulletproof'.  However, once I step away from the simulator and think about what can happen in the next 30 years, then things get interesting...

I would like to see more of this common sense discussion (like there are at E-R.org), I think folks are getting lost in the weeds calculating 'The Magic Number' (down to 3 sig figs) and not enough time discussing what that number means in the future.  Maybe step back and discuss why we actually think that number is good enough (like technology making retirement both more enjoyable and cheaper).

I also highlighted the idea that we are sitting on more valuable assets nowadays.  For instance, if I weren't so lazy, I could probably turn my 2 cars and a house into an income stream.

I would also want to hear more about if market history really is representative of the future.  I think there are great arguments why it isn't, and the only argument why it is is because 'we don't have anything else'.

BTW, thanks for the link to the Philosophical Economist - right up my alley!
Title: Re: Using market valuation measurements to affect SWR?
Post by: brooklynguy on May 08, 2015, 10:57:20 AM
I would like to see more of this common sense discussion (like there are at E-R.org), I think folks are getting lost in the weeds calculating 'The Magic Number' (down to 3 sig figs) and not enough time discussing what that number means in the future.  Maybe step back and discuss why we actually think that number is good enough (like technology making retirement both more enjoyable and cheaper).

. . .

I would also want to hear more about if market history really is representative of the future.  I think there are great arguments why it isn't, and the only argument why it is is because 'we don't have anything else'.

Maybe you're just not spending enough time in the forum...

Take a look at the discussion going on in this thread, for example:

http://forum.mrmoneymustache.com/welcome-to-the-forum/cfiresim-success-rate/
Title: Re: Using market valuation measurements to affect SWR?
Post by: EscapeVelocity2020 on May 08, 2015, 11:17:00 AM
Thanks BG, I'm not spending a lot of time on the forum, I'll admit, but I had seen the earlier part of that discussion.  I have to say, this bit drives me a bit crazy:
or the potential risks that history has yet to throw at us at all.

I'm fundamentally optimistic about the future.

I think the odds of another major terror attack (and subsequent revenge invasion) are significantly greater than zero, and I have a short list of personal favorites that I can't believe no one has tried yet.  Flying airplanes into buildings was a good one, but there are a bunch more ways that bad people could do major harm with limited resources.  I think most people don't realize how fragile civilization really is.

I wouldn't be surprised by a devastating outbreak of some antibiotic-resistant pathogen taking 10% of the world population in my lifetime.  Such diseases already exist, and new ones are evolving faster than our drugs are.  I don't think anyone believes our ability to scale up the production of new antibiotics can possible compete with exponentially growing infection numbers, so this is one of those deals where we rapidly lose control of if we don't stop it up front.  The planet has never had this many people living in such close quarters before, we're ripe for infection.

Natural disasters happen with some regularity.  What would the US economy do if Mt. Rainier wiped out half of Seattle?  If the Hayward fault took down half of Northern California?  Katrina was a speedbump in a second-tier city.

War.  Always war.  It's what humans do.  There are lots of ways for this to unfold next time, but my money is on some combination of the Middle East and China.

Climate Change.  In most places, changes will be slow and gradual and we'd have no problems adapting.  In some places, those slow changes will lead to very  rapid shifts.  Plate tectonics was slow, too, until South America suddenly disconnected from Antarctica and connected the Pacific and Atlantic oceans for the first time in a billion years and dramatically changed the planet's ocean circulation pattern.  That's the sort of problem with climate change I'm worried about.

Generic population growth.  There are currently more people living in southeast Asia than in the rest of the world combined.  I don't claim to understand what impact that will have on the future of the planet, but I don't believe it will be irrelevant to the next century.

And despite all of that, I still think the future of the world economy looks better than the past.  Standards of living are rising. The rule of law is expanding.  Life expectancy and literacy rates are climbing.  We've already survived a bunch of that bad stuff listed above, and we'll get through some more of it, but I still think the general trend for humanity (and the stock market) is up up and away for the remainder of my own short life.

Maybe I should go over to that thread to discuss it, but if I had all these horrible things on my radar screen as 'going to happen', SWR would be the last thing on my mind.  I'd be trying to make the most of what civilization has to offer right now since it's obviously going to be compromised in my lifetime :)

But hey, Sol makes me feel like I'm outrageously optimistic (and blissfully ignorant) because I don't actually think any of these things will happen or become a 'devastating event'!  I operate under the assumption that the worst 'global things' (like WWII) have already happened and the new 'global things' (like ebola) won't really affect me much.  Maybe not being on the forum is to my benefit...

Title: Re: Using market valuation measurements to affect SWR?
Post by: Dr. Doom on May 08, 2015, 03:23:06 PM
I think we're talking about three things here:

1) Are the US Domestic ETF markets overvalued?
Yes.  Period.  We can talk about "reasons" why they are overvalued until we're blue in the face, but at the end of the discussion, it's still a yes here.

2) Is CAPE a good measure of determining whether or not markets are overvalued?
Yes, particularly when it's *far* above norms.  Although the degree of underlying accuracy is in question, I think when CAPE is 27 and the historical average is below 20, it's fairly safe to say:  Absolutely. Things are looking a little high.  Maybe very high.

You could also examine Q ratio, price/book ratio, div yield on the S&P as other methods of determining overall valuation levels.  They all say:  High.

Will they remain high?  Maybe.  Maybe for a long time.  Maybe not for much longer.  You can't time the markets and none of these metrics should be used as an excuse for doing so.

There are two main knobs which will adjust as the market inevitably returns to its average valuation (or below.)

- Price reduction
- Earnings growth

Doesn't have to be a crash.  Might just be a slow growth in earnings over time.  Might be a slow erosion in price due to inflation or <other factors>.  Markets can potentially stay overvalued for extended periods of time.

3) Does retiring during periods when the market is overvalued change the way we should evaluate cFIREsim results?
Yes.

Look, by definition, the average of cFIREsim's runs start at an average market valuation.  Throughout history we can expect that exactly half of the start years for a 30 year retirement initiated at 'below average' valuations and the remaining half initiated at 'above average.'

But cFIREsim doesn't take that into account.  When it sorts through history, it just says:  OK, we're looking at all possible retirement terms available which satisfy that your term window.  So we can expect that it is analyzing portfolio performance when retiring against an average market valuation when taken in total.

Being that overvalued markets are more likely to 'correct' (read: regress to the mean) in the short term and therefore produce a bad initial sequence of returns (which is the greatest risk to your retirement success) then yes, you should be somewhat more concerned retiring when you know the market is overvalued.

In other words, cFIREsim is not telling the whole truth here when it spits out its success rate and I think there's value in being aware of this.

A bit more on retiring in an overvalued market:
I think given current valuations that retiring right now with a static 4% WR could be a dicey proposition if -- and this is a big if -- you are certain you cannot or will not ever ever (ever!) work again.  Dicey+ if you are a sole breadwinner, meaning: other people rely on you to live.  Dicey++ if you are unable to cut spending at all in tough times.

On the other hand, if your assets take a sharp tumble and you don't mind earning some additional income and perhaps combining that with "floor" spending, retiring today with a 4% withdrawal will work fine for you.  More than fine.  You will love the hell out of retiring immediately. Why haven't you retired already?

Just keep an eye on things, OK?

Also, to directly answer your question, I over-saved by some amount in order to combat my own concerns regarding current valuations.  I believe in statistics and regression to the mean.
Title: Re: Using market valuation measurements to affect SWR?
Post by: EscapeVelocity2020 on May 08, 2015, 09:32:08 PM
That's much more practical and useful discussion, IMHO, and there is a lot of interesting insight to be gained as opposed to trying to somehow adjust today's CAPE down to an equal comparison across market history.  CAPE can be sustained at higher values, but I still feel as though a lower SWR is warranted to guard against the high likelihood that the market is overvalued.  Just because the market failed only a hand full of times in the past doesn't matter if the present looks just as bad or worse than every time it failed.  The larger problem is that the Fed has an outsized influence on the market, so we haven't really been through this before to know how overvalued things are.

Some other interesting differences from the past are that inflation truly seems to be dead and unemployment is harder to fight for the Fed.

But I also get a lot more for my money, and have more options than the generations before us.  It seems possible that I spend less in the future (in real terms) because I have many more options to substitute goods and benefit from global competition.  That would also lower my WR, but not because I'm necessarily trying to keep it low.

So, if you over-saved, the worst cases are that you worked extra years and/or have an increased standard of living in the future.  But if you under-saved, you have to work more years in the future and/or have a declining standard of living.  I know which one I'm shooting for...
Title: Re: Using market valuation measurements to affect SWR?
Post by: brooklynguy on May 09, 2015, 09:03:37 AM
Look, by definition, the average of cFIREsim's runs start at an average market valuation.  Throughout history we can expect that exactly half of the start years for a 30 year retirement initiated at 'below average' valuations and the remaining half initiated at 'above average.'

These would be true statements if by "average" you mean "median", but in the context of your post (which refers to "regression to the mean") I don't think that's what you meant (and it's generally not what anyone means when they talk about the market being "overvalued").  In general, when someone says the market is "overvalued", they mean that the current value of their given valuation yardstick (like CAPE or any of the other metrics you named) is high in relation to that metric's historical mean, and they fear that the market will revert to the mean.  But, of course, the market has spent (and no doubt will continue to spend) periods of time (sometimes extended periods of time) in deviation from its mean, in a non-perfectly-symmetrical way, so it's not the case that the market has spent (or will spend) half the time "above average" and half the time "below average."

That said, I still generally agree with your overall point and think it would be wise to exercise caution in markets like the current one (personally I am "cautiously optimistic" about the future).

Quote
A bit more on retiring in an overvalued market:
I think given current valuations that retiring right now with a static 4% WR could be a dicey proposition if -- and this is a big if -- you are certain you cannot or will not ever ever (ever!) work again.  Dicey+ if you are a sole breadwinner, meaning: other people rely on you to live.  Dicey++ if you are unable to cut spending at all in tough times.

I generally agree with this bit too, but I would give greater weight to spending flexibility.  In my view, the primary reason history-based success rates are overly conservative is their unrealistic assumption that the retiree will robotically make the same fixed constant-dollar withdrawals year after year.
Title: Re: Using market valuation measurements to affect SWR?
Post by: Dr. Doom on May 09, 2015, 10:25:12 AM
In general, when someone says the market is "overvalued", they mean that the current value of their given valuation yardstick (like CAPE or any of the other metrics you named) is high in relation to that metric's historical mean, and they fear that the market will revert to the mean.
Right - you need to select a metric (such as CAPE or Q) in order to make a determination as to whether a given market is overvalued.

And part of this thread is discussing whether the historical means for these metrics are valid any longer -- or if the mean is artificially low, given how much things have changed in the past 100+ years.

I'm just making the point that when virtually all valuation metrics are showing the same thing (that markets appear to be valued much higher than historical norms) it seems extremely likely to me that the markets are, in fact, actually overpriced.

Quote
But, of course, the market has spent (and no doubt will continue to spend) periods of time (sometimes extended periods of time) in deviation from its mean, in a non-perfectly-symmetrical way, so it's not the case that the market has spent (or will spend) half the time "above average" and half the time "below average."
Yes, you're right - a sloppy statement on my part.  Another flaw in my initial statement:  Data excludes the last X years, where X is (retirement term - 1).

The central point remains though, which is that the history based calculators have no sense of valuations, and valuations do affect your chance of success.  Again, although we can have a discussion about the degree to which it matters, it's a fact that it matters.

Quote
(personally I am "cautiously optimistic" about the future).

Same.  Ideally you want to be aware of the risk, have backup plans (mine is called floor spending), and stay the course on your AA no matter what happens.  I'm not advocating that people start panicking, nor am I suggesting that everyone over-save.

Quote
I would give greater weight to spending flexibility.  In my view, the primary reason history-based success rates are overly conservative is their unrealistic assumption that the retiree will robotically make the same fixed constant-dollar withdrawals year after year.

Yes, I agree with this completely, and runs with cFIREsim bear this out -- spending less during downturns helps success rates enormously.  And in the real world, practically nobody blindly spends the same amount regardless of current conditions.

4% will be fine for the vast majority of people, even if they're retiring when indicators show that markets are somewhat high.

I will also add that Vanguard itself feels that markets are overvalued.
See video:
https://personal.vanguard.com/us/insights/video/3241-Exc1