A costly mistake made by both investors and advisors

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
larryswedroe
Posts: 16016
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

A costly mistake made by both investors and advisors

Post by larryswedroe » Tue Aug 02, 2016 7:32 am

http://www.advisorperspectives.com/arti ... dium=email

I have been surprised that there are advisors that use historical returns in financial planning.

Hope this is helpful in explaining why it's a mistake and right now may prove to be a particularly costly or dangerous one

Best wishes
Larry

User avatar
coachz
Posts: 1048
Joined: Wed Apr 04, 2007 7:10 am
Location: Charleston, SC

Re: A costly mistake made by both investors and advisors

Post by coachz » Tue Aug 02, 2016 7:42 am

Thank you as always for the thought provoking articles !!!

I'm not sure what qualifies for cherry picking dates but

1926 through 1948
1949 through 1999

seems like it to me. Is it valid to pick periods like this?

Topic Author
larryswedroe
Posts: 16016
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: A costly mistake made by both investors and advisors

Post by larryswedroe » Tue Aug 02, 2016 7:48 am

Coachz
Of course I cherry picked dates to make the point that valuations matter. Especially at extremes.

But as I have shown it doesn't matter where they are, valuations are the best metric we have to forecast future returns (with wide dispersion). Higher valuations forecasts lower means, worse worst cases and less good best cases. And rising valuations cause valuations to rise and thus lower future expected returns

I would note that in Europe the pension plans are required to use current valuations to project the liabilities. It's only in US that pension plans of government entities (not corporations) get to choose the forecast of returns they want.

Larry

User avatar
coachz
Posts: 1048
Joined: Wed Apr 04, 2007 7:10 am
Location: Charleston, SC

Re: A costly mistake made by both investors and advisors

Post by coachz » Tue Aug 02, 2016 7:51 am

Thanks Larry. I"m not understanding these two sentences. Would you be kind enough to break them down further please?
"Higher valuations forecasts lower means, worse worst cases and less good best cases. And rising valuations cause valuations to rise and thus lower future expected returns"

User avatar
in_reality
Posts: 4529
Joined: Fri Jul 12, 2013 6:13 am

Re: A costly mistake made by both investors and advisors

Post by in_reality » Tue Aug 02, 2016 7:51 am

coachz wrote:Thank you as always for the thought provoking articles !!!

I'm not sure what qualifies for cherry picking dates but

1926 through 1948
1949 through 1999

seems like it to me. Is it valid to pick periods like this?
No. Forget the past.
The best tool we have for forecasting equity returns are current valuations. With that in mind, the Shiller CAPE 10 earnings yields as of June 30, 2016 are 4.1% for the United States, 6.6% for developed markets and 8.2% for emerging markets.
Similarly, the research shows that today’s bond yields are a better predictor of future returns than are historical returns.

trueblueky
Posts: 1646
Joined: Tue May 27, 2014 3:50 pm

Re: A costly mistake made by both investors and advisors

Post by trueblueky » Tue Aug 02, 2016 7:53 am

larryswedroe wrote:http://www.advisorperspectives.com/arti ... dium=email

I have been surprised that there are advisors that use historical returns in financial planning.

Hope this is helpful in explaining why it's a mistake and right now may prove to be a particularly costly or dangerous one

Best wishes
Larry
Thank you. The negative correlation with historical was unexpected.

User avatar
coachz
Posts: 1048
Joined: Wed Apr 04, 2007 7:10 am
Location: Charleston, SC

Re: A costly mistake made by both investors and advisors

Post by coachz » Tue Aug 02, 2016 7:55 am

in_reality wrote:
coachz wrote:Thank you as always for the thought provoking articles !!!

I'm not sure what qualifies for cherry picking dates but

1926 through 1948
1949 through 1999

seems like it to me. Is it valid to pick periods like this?
No. Forget the past.
The best tool we have for forecasting equity returns are current valuations. With that in mind, the Shiller CAPE 10 earnings yields as of June 30, 2016 are 4.1% for the United States, 6.6% for developed markets and 8.2% for emerging markets.
Similarly, the research shows that today’s bond yields are a better predictor of future returns than are historical returns.
The past (maybe not in investing) is a great tool for me to learn from. I will never forget the past but I might find new ways to predict the future.

Grt2bOutdoors
Posts: 21725
Joined: Thu Apr 05, 2007 8:20 pm
Location: New York

Re: A costly mistake made by both investors and advisors

Post by Grt2bOutdoors » Tue Aug 02, 2016 8:08 am

Thanks for posting the article, always can rely on you for being candid. :beer
"One should invest based on their need, ability and willingness to take risk - Larry Swedroe" Asking Portfolio Questions

rbaldini
Posts: 1073
Joined: Mon Mar 23, 2015 3:20 pm

Re: A costly mistake made by both investors and advisors

Post by rbaldini » Tue Aug 02, 2016 10:31 am

Sincere question: if you can't use past data to estimate future returns (aided by good models and statistics), then what *can* you use? I'm not an economist; I don't know any data-free theories that would help me predict the future.

User avatar
coachz
Posts: 1048
Joined: Wed Apr 04, 2007 7:10 am
Location: Charleston, SC

Re: A costly mistake made by both investors and advisors

Post by coachz » Tue Aug 02, 2016 10:33 am

Larry,
Small typo on paragraph 5. I think "when the E/P is 3.4%" should be "when the E/P is 34.4%"

I read the entire article but it was over my football shaped head.

Also, this table might help others. In the future if you could make tables like this it would help to decode the paragraphs of numbers that seem to form a puzzle.

Image
Last edited by coachz on Tue Aug 02, 2016 10:37 am, edited 2 times in total.

RadAudit
Posts: 3645
Joined: Mon May 26, 2008 10:20 am
Location: Second star on the right and straight on 'til morning

Re: A costly mistake made by both investors and advisors

Post by RadAudit » Tue Aug 02, 2016 10:33 am

rbaldini wrote:Sincere question: if you can't use past data to estimate future returns (aided by good models and statistics), then what *can* you use?
Sincere answer. Read the article.
FI is the best revenge. LBYM. Invest the rest. Stay the course. - PS: The cavalry isn't coming, kids. You are on your own.

rbaldini
Posts: 1073
Joined: Mon Mar 23, 2015 3:20 pm

Re: A costly mistake made by both investors and advisors

Post by rbaldini » Tue Aug 02, 2016 10:36 am

I think I misread at first. Obviously past data is the only data we can use, and it is very helpful. The argument, I believe, is that past returns alone are not useful. But other kinds of data can be useful. Of course, useful statistics can only be discovered by looking at historical returns.

Topic Author
larryswedroe
Posts: 16016
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: A costly mistake made by both investors and advisors

Post by larryswedroe » Tue Aug 02, 2016 11:12 am

Coach
First to answer your first question about not understanding about best and worst cases, the answer is well laid out on page 3 in the bullet points discussing Asness's work
http://www.advisorperspectives.com/arti ... ors-make/3

Also E/P cannot be 34.4 unless P/E is under 3, never happened.

Re using past data, Clearly we can learn a lot from historical data, including things like correlations and volatility and also returns. But one has to understand HOW the returns were earned. In other words if the returns were earned in repeatable way they have more value. But if returns are earned by rising valuations that's another story. So if bond yields fall, that drives up past/historical returns to bonds, but clearly now future expected returns are lower. Same thing for stocks. Doesn't mean they WILL be lower, but it means they likely will be UNLESS valuations rise further. And at some point we know "trees don't grow to the sky." So if the current P/E or CAPE 10 was at historical average and bond yields about their historical average you could use historical returns data with a lot more confidence.

I hope that helps
Larry
Last edited by larryswedroe on Tue Aug 02, 2016 11:17 am, edited 1 time in total.

soboggled
Posts: 901
Joined: Mon Jun 27, 2016 10:26 am

Re: A costly mistake made by both investors and advisors

Post by soboggled » Tue Aug 02, 2016 11:16 am

Next up: The fallacy of out-of-context backtesting.

User avatar
coachz
Posts: 1048
Joined: Wed Apr 04, 2007 7:10 am
Location: Charleston, SC

Re: A costly mistake made by both investors and advisors

Post by coachz » Tue Aug 02, 2016 11:51 am

larryswedroe wrote:Coach
First to answer your first question about not understanding about best and worst cases, the answer is well laid out on page 3 in the bullet points discussing Asness's work
http://www.advisorperspectives.com/arti ... ors-make/3

Also E/P cannot be 34.4 unless P/E is under 3, never happened.

Re using past data, Clearly we can learn a lot from historical data, including things like correlations and volatility and also returns. But one has to understand HOW the returns were earned. In other words if the returns were earned in repeatable way they have more value. But if returns are earned by rising valuations that's another story. So if bond yields fall, that drives up past/historical returns to bonds, but clearly now future expected returns are lower. Same thing for stocks. Doesn't mean they WILL be lower, but it means they likely will be UNLESS valuations rise further. And at some point we know "trees don't grow to the sky." So if the current P/E or CAPE 10 was at historical average and bond yields about their historical average you could use historical returns data with a lot more confidence.

I hope that helps
Larry

Thanks Larry. My mistake. I'm gonna have to work till I'm dead with this market. I'll be at the river living in my van eating government cheese. :sharebeer

User avatar
siamond
Posts: 5125
Joined: Mon May 28, 2012 5:50 am

Re: A costly mistake made by both investors and advisors

Post by siamond » Tue Aug 02, 2016 12:18 pm

It would have been very useful to clarify that expected returns forecasts (with all the uncertainty coming with it) are about the coming 10 to 15 years, and NOT about the rest of our lives. Whether we're accumulators or retirees, this makes a HUGE difference, as periods of low returns are typically followed by periods of high returns. Today is not a fun situation for sure, but the US history includes multiple years that were probably much worse starting points. And yet the returns over multiple decades were pretty good, because return-to-the-mean goes both ways.

And then this puts the following key statement of the article in context:
Specifically, it’s the error of using historical stock and bond returns as the best estimator of their future returns.

If you mean by 'future returns' the expected returns of the coming decade, yes, this statement is probably largely true. But most casual readers will interpret this statement as "historical data is completely irrelevant, for my entire retirement planning", and this couldn't be more misguided. I wish this would have been made more clear.

bigred77
Posts: 2028
Joined: Sat Jun 11, 2011 4:53 pm

Re: A costly mistake made by both investors and advisors

Post by bigred77 » Tue Aug 02, 2016 12:52 pm

siamond wrote:It would have been very useful to clarify that expected returns forecasts (with all the uncertainty coming with it) are about the coming 10 to 15 years, and NOT about the rest of our lives. Whether we're accumulators or retirees, this makes a HUGE difference, as periods of low returns are typically followed by periods of high returns. Today is not a fun situation for sure, but the US history includes multiple years that were probably much worse starting points. And yet the returns over multiple decades were pretty good, because return-to-the-mean goes both ways.

And then this puts the following key statement of the article in context:
Specifically, it’s the error of using historical stock and bond returns as the best estimator of their future returns.

If you mean by 'future returns' the expected returns of the coming decade, yes, this statement is probably largely true. But most casual readers will interpret this statement as "historical data is completely irrelevant, for my entire retirement planning", and this couldn't be more misguided. I wish this would have been made more clear.
+1...

I agree with the article and thought it was very well written. I just think the the following points need to be stressed (keeping in mind this is my understanding of the subject in laymen terms. I don't claim to be anywhere near the ballpark that Larry is in):

- Using CAPE 10 is probably the best tool we have to derive expected returns. However, that doesn't make it an objectively "good" tool. Research shows it explains roughly 50% of actual returns observed over the subsequent 10 year period. It's simply the best we have at our disposal.

- Valuations do matter. They matter the most in estimating returns over intermediate time periods (lets say a decade or 2 for simplicity). They matter less in estimating returns over the long term (multiple decades, half a century, ones lifetime investing time frame).They don't matter at all in the very short term (what will the market do tomorrow).


The following is a quote I took directly from Larry's paper (I hope that's OK)... "That results in expected real returns to U.S. stocks of 4.5%, to developed market stocks of 7.3% and to emerging market stocks of 9.0%. To get an estimate of nominal returns, you can add the current spread between the 10-year nominal bond and the 10-year TIPS, which is currently about 1.5%." ... Those estimations seem entirely reasonable to me. Fantastic. Why then is there so much debate and hand wringing here lately about SWRs and worst case scenarios? If current valuations are implying these expected returns then why are we so concerned about conservative investors not being conservative enough? Expected real returns for equity portions of portfolios are going to be roughly 6% if you have a 50/50 mix of US TSM and Total International.

User avatar
coachz
Posts: 1048
Joined: Wed Apr 04, 2007 7:10 am
Location: Charleston, SC

Re: A costly mistake made by both investors and advisors

Post by coachz » Tue Aug 02, 2016 12:57 pm

bigred77 wrote:Expected real returns for equity portions of portfolios are going to be roughly 6% if you have a 50/50 mix of US TSM and Total International.
6% real supports 4% SWR so what's the problem ? I feel dumber with every post ! :idea:

Topic Author
larryswedroe
Posts: 16016
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: A costly mistake made by both investors and advisors

Post by larryswedroe » Tue Aug 02, 2016 1:04 pm

It would have been very useful to clari ... our lives.

No, this is incorrect. When financial economists forecast returns they are UNCONDITIONAL returns, meaning regardless of the term.

Same applies to bigred's comment. The forecasts apply to long term, like 30 years, as well,

Re 6%, If US real is about 4, and developed about 7 and EM about 9, a market cap weighted would get you near 6, but that is only the equity portion. And that is if you do diversify as much internationally. Remember the average US investor has only about 10% international.

Then of course you have the bond side with say 0 real. So a 50/50 gets IF internationally diversified gets you 3% real, which clearly isn't going to support a 4% SWR.

Larry

freyj6
Posts: 216
Joined: Thu Jun 19, 2014 2:09 am

Re: A costly mistake made by both investors and advisors

Post by freyj6 » Tue Aug 02, 2016 1:10 pm

This is more or less what hit me really hard when reading Tony Robbins' book last year.

He shows how a risk parity portfolio with 40% long bonds, 15% shorter bonds, 7.5% gold, 7.5% commodities and 30% stocks (I think) did really great... in retrospect.

Now that bond yields are virtually at all time lows, insuring commodities seems to have lost its premium and gold is well above its historical average... PERHAPS we won't see a repeat lol.

bigred77
Posts: 2028
Joined: Sat Jun 11, 2011 4:53 pm

Re: A costly mistake made by both investors and advisors

Post by bigred77 » Tue Aug 02, 2016 1:11 pm

coachz wrote:
bigred77 wrote:Expected real returns for equity portions of portfolios are going to be roughly 6% if you have a 50/50 mix of US TSM and Total International.
6% real supports 4% SWR so what's the problem ? I feel dumber with every post ! :idea:
It does, but that 6% is strictly for the equity portion of the portfolio. A reasonable portfolio for a retiree is going to have a significant fixed income component, with expected returns a heck of a lot lower (for the specific component of the portfolio).

However, a retiree really only needs their portfolio to achieve a touch above a 1% real return for a 4% withdrawal rate to last 30 years (not taking into account sequence of returns risk).

User avatar
coachz
Posts: 1048
Joined: Wed Apr 04, 2007 7:10 am
Location: Charleston, SC

Re: A costly mistake made by both investors and advisors

Post by coachz » Tue Aug 02, 2016 1:14 pm

Is this true?

"a retiree really only needs their portfolio to achieve a touch above a 1% real return for a 4% withdrawal rate to last 30 years".

I can lock in CDs and get that. Plus I have no bequests I want to spend it down. When you say last 30 years does that include spending it down ? I would guess so.

freyj6
Posts: 216
Joined: Thu Jun 19, 2014 2:09 am

Re: A costly mistake made by both investors and advisors

Post by freyj6 » Tue Aug 02, 2016 1:24 pm

Coachz, it depends on the sequence of returns.

If you had a stead 1% a year all the way through, or bigger gains in the first half of retirement, you could easily fund a retirement like that.

However, if you had negative returns early and positive returns late, you'd likely run out of money.

WasabiOsbourne
Posts: 308
Joined: Wed Mar 16, 2016 1:48 pm

Re: A costly mistake made by both investors and advisors

Post by WasabiOsbourne » Tue Aug 02, 2016 1:31 pm

larry, thanks... good stuff :)

prime example is Tony Robbins book talking about wonderful all-seasons portfolio of Ray Dalio. amazingly positive returns with limited drawdowns. but of course the heavy government bond weighting isn't going to perform like it did in the past. i'm sure Ray is aware of this and points it out. but Tony is going with it - wonder if anyone pointed it out to him before his book

FWIW, i like Tony and his book in general.

EDIT: someone beat me to tony robbins it seems...
Last edited by WasabiOsbourne on Tue Aug 02, 2016 1:34 pm, edited 1 time in total.

User avatar
coachz
Posts: 1048
Joined: Wed Apr 04, 2007 7:10 am
Location: Charleston, SC

Re: A costly mistake made by both investors and advisors

Post by coachz » Tue Aug 02, 2016 1:32 pm

WasabiOsbourne wrote:larry, thanks... good stuff :)

prime example is Tony Robbins book talking about wonderful all-seasons portfolio of Ray Dalio. amazingly positive returns with limited drawdowns. but of course the heavy government bond weighting isn't going to perform like it did in the past. i'm sure Ray is aware of this and points it out. but Tony is going with it - wonder if anyone pointed it out to him before his book

FWIW, i like Tony and his book in general.
Just don't let Tony talk you into walking on coals.

Topic Author
larryswedroe
Posts: 16016
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: A costly mistake made by both investors and advisors

Post by larryswedroe » Tue Aug 02, 2016 1:49 pm

Try this as one example I ran in past (doing this from memory) From 1973 onward stocks returned 7% real, yet if you withdrew 7% real if memory serves you would have been bankrupt in under a decade, the order of returns matters.
Larry

NMJack
Posts: 836
Joined: Sun Feb 14, 2016 1:22 pm

Re: A costly mistake made by both investors and advisors

Post by NMJack » Tue Aug 02, 2016 2:07 pm

Larry: I'm trying to wrap my head around your using E/P as the sole basis for forecasting future returns. Four hypothetical companies:

All have a market cap or $1B, one million shares outstanding (i.e. $1000 per share price), corporate earnings of $50M and thus P/E of 20. Using your model, the E/P would be .05, which you say would project a 5% ROI.

Company A:
Has $200,000,000 of cash on the books doing nothing. They decide to implement a stock buyback plan, and buy back 200,000 shares of stock at the current price of $1000 per share. There are now 800,000 shares of stock outstanding. This increases the earnings per share from $50 to $62.50, reducing the P/E to 16. Now, granted, during the share buyback, the price per share would likely increase, but that would reward the shareholder in a similar manner. Does your model now increase the ROI from 5% to 6.25%?

Company B:
Also has $200,000,000 cash on the books, but decides to aggressively spend on R&D that will have less than a 3 year payback (so will increase earnings by at least 33% of the $200,000,000 spent per year). So, annual earnings go up to $50M + $200M x .33 = $116M. That changes the P/E from 20 to 9. Does your model now increase ROI from 5% to 11%?

Company C:
Has $100 on the books, so just continues along as usual.

Company D:
Highly leveraged. Has outstanding debt of $200,000,000 but is still paying dividends and at times borrowing money to do so.

I'm no accountant (and I can prove it), so I may not understand any of this. If not, I would appreciate more knowledgeable individuals "kindly" educating me. That said, I believe that there are many more variables on a corporate balance sheet that need to be considered beyond P/E ratio.

Thanks.

NMJack
Posts: 836
Joined: Sun Feb 14, 2016 1:22 pm

Re: A costly mistake made by both investors and advisors

Post by NMJack » Tue Aug 02, 2016 2:50 pm

NMJack wrote:Larry: I'm trying to wrap my head around your using E/P as the sole basis for forecasting future returns. Four hypothetical companies:

All have a market cap or $1B, one million shares outstanding (i.e. $1000 per share price), corporate earnings of $50M and thus P/E of 20. Using your model, the E/P would be .05, which you say would project a 5% ROI.

Company A:
Has $200,000,000 of cash on the books doing nothing. They decide to implement a stock buyback plan, and buy back 200,000 shares of stock at the current price of $1000 per share. There are now 800,000 shares of stock outstanding. This increases the earnings per share from $50 to $62.50, reducing the P/E to 16. Now, granted, during the share buyback, the price per share would likely increase, but that would reward the shareholder in a similar manner. Does your model now increase the ROI from 5% to 6.25%?

Company B:
Also has $200,000,000 cash on the books, but decides to aggressively spend on R&D that will have less than a 3 year payback (so will increase earnings by at least 33% of the $200,000,000 spent per year). So, annual earnings go up to $50M + $200M x .33 = $116M. That changes the P/E from 20 to 9. Does your model now increase ROI from 5% to 11%?

Company C:
Has $100 on the books, so just continues along as usual.

Company D:
Highly leveraged. Has outstanding debt of $200,000,000 but is still paying dividends and at times borrowing money to do so.

I'm no accountant (and I can prove it), so I may not understand any of this. If not, I would appreciate more knowledgeable individuals "kindly" educating me. That said, I believe that there are many more variables on a corporate balance sheet that need to be considered beyond P/E ratio.

Thanks.
I will postulate that the above will not occur in the real world. Despite the fact that all mathematically would generate the same P/E (and thus E/P), there are far different inherent values, and thus potential future returns, in each case. The investment world would recognize this and drive up the stock prices of Companies A and B and drive down the stock prices of C and D. A and B might be selling at $1250 per share (P/E = 25) while C might be selling at $750 (P/E = 15) and D might be selling at $500 (P/E = 10).

Is it possible that in the early years cited in Larry's research that companies operated more like company C and company D? I believe that in the decades following the great depression, companies did not accumulate cash and paid large dividends (likely in large part due to the mistrust of institutions in general after the depression). I believe that in today's environment, more companies operate like A and B. We read countless reports of record balance sheets and aggressive stock buy backs. Could this explain much of the high current P/E ratios, and thus shoot holes in using historical vs. current E/P to make forecasts regarding future real earnings? I'm just asking.... :confused

Topic Author
larryswedroe
Posts: 16016
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: A costly mistake made by both investors and advisors

Post by larryswedroe » Tue Aug 02, 2016 2:54 pm

NMJack
Simple when companies buy back stock it changes lots of things including their financial strength, leverage perhaps, EPS, etc. The market adjusts to the new information and the E/P changes accordingly. Just like when any other new information comes along
Larry

NMJack
Posts: 836
Joined: Sun Feb 14, 2016 1:22 pm

Re: A costly mistake made by both investors and advisors

Post by NMJack » Tue Aug 02, 2016 3:11 pm

larryswedroe wrote:NMJack
Simple when companies buy back stock it changes lots of things including their financial strength, leverage perhaps, EPS, etc. The market adjusts to the new information and the E/P changes accordingly. Just like when any other new information comes along
Larry
I understand all that. I'm talking about before they buy the stock back, but have the large amount of cash on their books that "could" allow that. That is when the stock price goes up, not when they announce the buyback. Announcing (and executing) the buyback is all after the fact. It is already built into the stock price (hence making a higher P/E "legit," vs risk).

I think the use of a single element of a company's overall balance sheet (i.e. the P/E ratio) is extremely simplistic, leaves out many other very important variables, and is highly likely to lead to useless forecasts. (see my last two posts for further explanation)

Topic Author
larryswedroe
Posts: 16016
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: A costly mistake made by both investors and advisors

Post by larryswedroe » Tue Aug 02, 2016 3:31 pm

NMJACK
Well the data shows you can also use p/cf or ebitda as well, but it's most available for p/e and fwiw, the academic research, see Damodaran, has found it to be the best predictor.
And stocks often go up when they announce buy backs or dividend increases or other similar actions.
Larry

bigred77
Posts: 2028
Joined: Sat Jun 11, 2011 4:53 pm

Re: A costly mistake made by both investors and advisors

Post by bigred77 » Tue Aug 02, 2016 4:35 pm

larryswedroe wrote:It would have been very useful to clari ... our lives.

No, this is incorrect. When financial economists forecast returns they are UNCONDITIONAL returns, meaning regardless of the term.

Same applies to bigred's comment. The forecasts apply to long term, like 30 years, as well,

Re 6%, If US real is about 4, and developed about 7 and EM about 9, a market cap weighted would get you near 6, but that is only the equity portion. And that is if you do diversify as much internationally. Remember the average US investor has only about 10% international.

Then of course you have the bond side with say 0 real. So a 50/50 gets IF internationally diversified gets you 3% real, which clearly isn't going to support a 4% SWR.

Larry
Larry,

While financial economists may attempt to forecast unconditional returns with an undefined term, the correlation between expected returns and actual observed returns varies significantly over the length of time frames we might try to measure.

I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over a 3 day time frame are pretty close to 0. I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over an 10 year time frame are pretty close to 0.5 (at least I'd accept that statement as reasonably accurate). I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over a 30 year time frame are somewhere between 0 and 0.5. I'd guess its lower than 0.25 but I don't have any exact data.

My point is I think valuations are at least useful in estimating what future returns might look like over the next 10-15 years. If my time horizon is 30+ years, I don't think expected returns derived from using CAPE 10 help me all that much.

NMJack
Posts: 836
Joined: Sun Feb 14, 2016 1:22 pm

Re: A costly mistake made by both investors and advisors

Post by NMJack » Tue Aug 02, 2016 4:40 pm

How many of the historical forecasts were made against a backdrop of huge corporate balance sheets and ZIRP? If the answer in "none," or "close to none," then I struggle to see how you can use the same approach that worked in back testing against one historical backdrop when today's backdrop is radically different. I'm not suggesting that relative P/E is not useful today, it's just when you compare today's P/E with P/E's in a different era, when corporations operated differently, I don't agree with the assumption that forward looking forecasts will behave in the same manner.

Simplistically:

When corporations paid out most or all of their earnings (and even borrowed money to pay dividends in recessions), it seems somewhat straightforward to use E/P on an absolute basis to forecast future returns.

Now, when corporations have been retaining earnings and building huge balance sheets (enabling smart growth investments, future stock buybacks, etc.), E/P on an absolute basis will understate future returns as it ignores the opportunity of immediate access to on-hand (cheap and excess) capital.

Again, if somebody who is in corporate accounting disagrees, please weigh in. I get it that the academic types have written books on this stuff, but like the saying goes, "those who can...do, those who can't...teach" and the corollary, "those who can't teach...write books." I first heard that one nearly 35 years ago, and after years in University and Megacorp, I really gained an appreciation for it.

soboggled
Posts: 901
Joined: Mon Jun 27, 2016 10:26 am

Re: A costly mistake made by both investors and advisors

Post by soboggled » Tue Aug 02, 2016 4:44 pm

bigred77 wrote:
larryswedroe wrote:It would have been very useful to clari ... our lives.

No, this is incorrect. When financial economists forecast returns they are UNCONDITIONAL returns, meaning regardless of the term.

Same applies to bigred's comment. The forecasts apply to long term, like 30 years, as well,

Re 6%, If US real is about 4, and developed about 7 and EM about 9, a market cap weighted would get you near 6, but that is only the equity portion. And that is if you do diversify as much internationally. Remember the average US investor has only about 10% international.

Then of course you have the bond side with say 0 real. So a 50/50 gets IF internationally diversified gets you 3% real, which clearly isn't going to support a 4% SWR.

Larry
Larry,

While financial economists may attempt to forecast unconditional returns with an undefined term, the correlation between expected returns and actual observed returns varies significantly over the length of time frames we might try to measure.

I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over a 3 day time frame are pretty close to 0. I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over an 10 year time frame are pretty close to 0.5 (at least I'd accept that statement as reasonably accurate). I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over a 30 year time frame are somewhere between 0 and 0.5. I'd guess its lower than 0.25 but I don't have any exact data.

My point is I think valuations are at least useful in estimating what future returns might look like over the next 10-15 years. If my time horizon is 30+ years, I don't think expected returns derived from using CAPE 10 help me all that much.
1. Not everyone's time frame is 30 years.
2. It may make sense to be a bit more conservative or a bit more aggressive at times based on expected returns.
3. Where did you get the .25 correlation? It takes quite awhile to overcome 10 years of suppressed returns and low returns in early retirement can hurt badly. As capital diminishes, a 10% return on $200K in later years is small consolation for 0% on $1M in earlier years..

User avatar
HomerJ
Posts: 13540
Joined: Fri Jun 06, 2008 12:50 pm

Re: A costly mistake made by both investors and advisors

Post by HomerJ » Tue Aug 02, 2016 4:53 pm

larryswedroe wrote:
It would have been very useful to clarify that expected returns forecasts (with all the uncertainty coming with it) are about the coming 10 to 15 years, and NOT about the rest of our lives.
No, this is incorrect. When financial economists forecast returns they are UNCONDITIONAL returns, meaning regardless of the term.
You need to explain this further. I'm guessing this is some economist jargon that has a special meaning for them. But then you are using it here without explaining it. Why?

Because it can't mean what it looks like you are saying. Ecomomists are taking today's valuations, and making a prediction that returns will be low regardless of term? And you are claiming this prediction is accurate, regardless of term? And actionable, regardless of term?
Then of course you have the bond side with say 0 real. So a 50/50 gets IF internationally diversified gets you 3% real, which clearly isn't going to support a 4% SWR.
3% real returns will definitely support a 30-year retirement where you are pulling out 4% real. The math is easy. Now sequence of return risk is real, but if one got a steady 3% real, of course one could pull 4% real and not run out of money. You will have LESS then when you started, but you won't run out.

Maybe that's why we've been talking past each other. Do you think SWR means ending up with more money than you started? SWR to most people means not going to zero in 30 years.
Last edited by HomerJ on Tue Aug 02, 2016 5:02 pm, edited 3 times in total.

User avatar
Artsdoctor
Posts: 4186
Joined: Thu Jun 28, 2012 3:09 pm
Location: Los Angeles, CA

Re: A costly mistake made by both investors and advisors

Post by Artsdoctor » Tue Aug 02, 2016 4:55 pm

Terrific article as always, Larry. One point, though:

The article is directed to individual investors and advisors. Don't you think it would be better to entitle it, "Very costly mistakes that pension fund managers are making"? THEIR projections are ridiculous!

User avatar
HomerJ
Posts: 13540
Joined: Fri Jun 06, 2008 12:50 pm

Re: A costly mistake made by both investors and advisors

Post by HomerJ » Tue Aug 02, 2016 4:55 pm

bigred77 wrote:
larryswedroe wrote:It would have been very useful to clari ... our lives.

No, this is incorrect. When financial economists forecast returns they are UNCONDITIONAL returns, meaning regardless of the term.

Same applies to bigred's comment. The forecasts apply to long term, like 30 years, as well,

Re 6%, If US real is about 4, and developed about 7 and EM about 9, a market cap weighted would get you near 6, but that is only the equity portion. And that is if you do diversify as much internationally. Remember the average US investor has only about 10% international.

Then of course you have the bond side with say 0 real. So a 50/50 gets IF internationally diversified gets you 3% real, which clearly isn't going to support a 4% SWR.

Larry
Larry,

While financial economists may attempt to forecast unconditional returns with an undefined term, the correlation between expected returns and actual observed returns varies significantly over the length of time frames we might try to measure.

I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over a 3 day time frame are pretty close to 0. I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over an 10 year time frame are pretty close to 0.5 (at least I'd accept that statement as reasonably accurate). I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over a 30 year time frame are somewhere between 0 and 0.5. I'd guess its lower than 0.25 but I don't have any exact data.

My point is I think valuations are at least useful in estimating what future returns might look like over the next 10-15 years. If my time horizon is 30+ years, I don't think expected returns derived from using CAPE 10 help me all that much.
This... Good post. The distinctions ARE important. CAPE proponents like to throw out the 40% (or 50%) correlation numbers (which is still pretty darn low), but that's not for ALL time periods, and it's very disingenuous to pretend that valuations are useful, regardless of term.

bigred77
Posts: 2028
Joined: Sat Jun 11, 2011 4:53 pm

Re: A costly mistake made by both investors and advisors

Post by bigred77 » Tue Aug 02, 2016 5:01 pm

soboggled wrote:
bigred77 wrote:
larryswedroe wrote:It would have been very useful to clari ... our lives.

No, this is incorrect. When financial economists forecast returns they are UNCONDITIONAL returns, meaning regardless of the term.

Same applies to bigred's comment. The forecasts apply to long term, like 30 years, as well,

Re 6%, If US real is about 4, and developed about 7 and EM about 9, a market cap weighted would get you near 6, but that is only the equity portion. And that is if you do diversify as much internationally. Remember the average US investor has only about 10% international.

Then of course you have the bond side with say 0 real. So a 50/50 gets IF internationally diversified gets you 3% real, which clearly isn't going to support a 4% SWR.

Larry
Larry,

While financial economists may attempt to forecast unconditional returns with an undefined term, the correlation between expected returns and actual observed returns varies significantly over the length of time frames we might try to measure.

I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over a 3 day time frame are pretty close to 0. I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over an 10 year time frame are pretty close to 0.5 (at least I'd accept that statement as reasonably accurate). I would argue that the correlation between expected returns derived from using the CAPE 10 and the actual returns I would observe over a 30 year time frame are somewhere between 0 and 0.5. I'd guess its lower than 0.25 but I don't have any exact data.

My point is I think valuations are at least useful in estimating what future returns might look like over the next 10-15 years. If my time horizon is 30+ years, I don't think expected returns derived from using CAPE 10 help me all that much.
1. Not everyone's time frame is 30 years. I agree completely. Not everyone's time frame is 10 years either.
2. It may make sense to be a bit more conservative or a bit more aggressive at times based on expected returns. You can make that argument if you want but I would never advise anyone to alter their AA based strictly on valuations or expected returns.
3. Where did you get the .25 correlation? It takes quite awhile to overcome 10 years of suppressed returns and low returns in early retirement can hurt badly. As capital diminishes, a 10% return on $200K in later years is small consolation for 0% on $1M in earlier years.. That 0.25 correlation is completely my top of the head guess. The only concrete assertion I was trying to make is that the correlations between expected and realized returns peak over the intermediate term and are lower over extended periods of time (multiple decades).

I would also point out that many posters are referencing sequence of return risk. To my knowledge, the CAPE 10 figure can tell us nothing useful about future volatility or sequence of returns. It may be useful to derive 4.5% expected real returns for US equities, but it tells of nothing about the likelihood of those returns looking like this (4.5%, 4.5%, 4.5%, 4.5%...) or like this (9%, -25%, 0%, 46%...)

Topic Author
larryswedroe
Posts: 16016
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: A costly mistake made by both investors and advisors

Post by larryswedroe » Tue Aug 02, 2016 5:05 pm

Homer, not responding any longer because it's fruitless
You completely misinterpret things, like 3% real return and assume it's not time varying. Doesn't mean you get 3% every year. In fact I explained you run and MCS based on the real return and the volatility of that return.

Soboggled, Of course if horizon is shorter you can take higher SWR, that's obvious. Foolish say for a 90 year old to have a 4% withdrawal rate.

BigRed
yes of course CAPE 10, or any other metric has less value over short periods. So does the DDM or any other metric.
But the math is what matters. It simply assumes valuations don't change because we cannot forecast if they will change. So earnings yield is like bond yield. You buy a 10 year bond and in short term doesn't matter what valuations do at end of 10 years you get the bond yield (assuming no default). Same with DDM or CAPE 10, which is nothing more than earnings yield. If valuations dont change you get in case of DDM the current div yield plus G of div (earnings which in long run are the same). It's the speculative return that is the big issue in short and longer term. But we cannot forecast it. What we do know is that stocks don't grow to sky. So unless you forecast higher than historical earnings growth (no reason i can think of to do that) the term should not matter. The mean expected return should be the same

And CAPE 10 doesn't tell you anything about sequence risk, that is why you run MCS to look at thousands of possible draws (and different sequences).

And I very specifically stated one should consider multiple factors before changing AA, including ability, willingness and need to take risks (and options one can exercise) but NEED is determined by expected returns which are determined by current valuations.

That's my last post on this subject as just repeating the same points.

User avatar
HomerJ
Posts: 13540
Joined: Fri Jun 06, 2008 12:50 pm

Re: A costly mistake made by both investors and advisors

Post by HomerJ » Tue Aug 02, 2016 5:21 pm

larryswedroe wrote:But the math is what matters. It simply assumes valuations don't change because we cannot forecast if they will change.
Horrible assumption with no basis in reality that destroys the entire model.

NMJack
Posts: 836
Joined: Sun Feb 14, 2016 1:22 pm

Re: A costly mistake made by both investors and advisors

Post by NMJack » Tue Aug 02, 2016 5:27 pm

larryswedroe wrote: That's my last post on this subject as just repeating the same points.
Does that mean that you're not planning to respond to my (hopefully well though out) questions regarding the impact of huge US corporate balance sheets and ZIRP (both historical anomalies) on using historical E/P patterns to forecast future returns?

Again, hopefully somebody with some Megacorp accounting credentials will weigh in on those. :sharebeer

Da5id
Posts: 2119
Joined: Fri Feb 26, 2016 8:20 am

Re: A costly mistake made by both investors and advisors

Post by Da5id » Tue Aug 02, 2016 5:28 pm

HomerJ wrote:
larryswedroe wrote:But the math is what matters. It simply assumes valuations don't change because we cannot forecast if they will change.
Horrible assumption with no basis in reality that destroys the entire model.
What assumption would you make about earnings going forward if not the same? Do you think earnings are about to take off? Why are his assumptions horrible?

I don't know, sometimes it seems like the 4% rule is a religious belief, and to challenge it is heresy. The question is, apparently, whether Larry is heavier than a duck...

bigred77
Posts: 2028
Joined: Sat Jun 11, 2011 4:53 pm

Re: A costly mistake made by both investors and advisors

Post by bigred77 » Tue Aug 02, 2016 5:29 pm

larryswedroe wrote:
BigRed
yes of course CAPE 10, or any other metric has less value over short periods. So does the DDM or any other metric.
But the math is what matters. It simply assumes valuations don't change because we cannot forecast if they will change. So earnings yield is like bond yield. You buy a 10 year bond and in short term doesn't matter what valuations do at end of 10 years you get the bond yield (assuming no default). Same with DDM or CAPE 10, which is nothing more than earnings yield. If valuations dont change you get in case of DDM the current div yield plus G of div (earnings which in long run are the same). It's the speculative return that is the big issue in short and longer term. But we cannot forecast it. What we do know is that stocks don't grow to sky. So unless you forecast higher than historical earnings growth (no reason i can think of to do that) the term should not matter. The mean expected return should be the same

And CAPE 10 doesn't tell you anything about sequence risk, that is why you run MCS to look at thousands of possible draws (and different sequences).

And I very specifically stated one should consider multiple factors before changing AA, including ability, willingness and need to take risks (and options one can exercise) but NEED is determined by expected returns which are determined by current valuations.

That's my last post on this subject as just repeating the same points.
I agree with the entirety of your response 100%.

The only thing I was trying to point out (possibly poorly) is that over really long time horizons, expected returns derived from valuations are have less predictive power in relation to realized returns. The reason being exactly what you pointed out: greater uncertainty in the earnings growth rate, our inability to forecast speculative return, etc. Over long horizons, even small changes to these variables have big consequences.

I appreciate you engaging on this topic.

NMJack
Posts: 836
Joined: Sun Feb 14, 2016 1:22 pm

Re: A costly mistake made by both investors and advisors

Post by NMJack » Tue Aug 02, 2016 5:34 pm

Da5id wrote: I don't know, sometimes it seems like the 4% rule is a religious belief, and to challenge it is heresy. The question is, apparently, whether Larry is heavier than a duck...
Isn't the question "does he weigh the same as a duck?" You know, duck floats in water, wood also floats in water, wood burns.... etc. (thanks for the much needed comic relief) :sharebeer

User avatar
HomerJ
Posts: 13540
Joined: Fri Jun 06, 2008 12:50 pm

Re: A costly mistake made by both investors and advisors

Post by HomerJ » Tue Aug 02, 2016 5:47 pm

Da5id wrote:
HomerJ wrote:
larryswedroe wrote:But the math is what matters. It simply assumes valuations don't change because we cannot forecast if they will change.
Horrible assumption with no basis in reality that destroys the entire model.
What assumption would you make about earnings going forward if not the same? Do you think earnings are about to take off? Why are his assumptions horrible?

I don't know, sometimes it seems like the 4% rule is a religious belief, and to challenge it is heresy. The question is, apparently, whether Larry is heavier than a duck...
He said the expected return based on valuations is UNCONDITIONAL.

4.1% expected return over the next 1 year
4.1% expected return over the next 3 years
4.1% expected return over the next 10 years
4.1% expected return over the next 20 years
4.1% expected return over the next 30 years
4.1% expected return over the next 50 years

We ask "How can all these be accurate?"

The answer, "Oh, we assume valuations won't change"

But next year, they will have new "expected" returns to publish. Unconditionally. So useful.

User avatar
Tyler9000
Posts: 476
Joined: Fri Aug 21, 2015 11:57 am

Re: A costly mistake made by both investors and advisors

Post by Tyler9000 » Tue Aug 02, 2016 5:48 pm

I see Larry's point that looking at a single historical timeframe for the purpose of predicting the future is likely to deceive, but by identifying a timeframe with similar economic conditions one can generate a better estimate. That definitely makes sense, but I would stop short of calling the use of historical data a mistake. The correlation of CAPE 10 to forward-looking returns promoted in the article is also based on studying historical returns, after all. :wink:

In general, I personally believe investors are better off truly understanding the inherent uncertainty of a portfolio choice rather than trying to pin down a specific future outcome with precision. Not all portfolios are equally uncertain, and studying history can be quite helpful in identifying one robust enough to handle any outcome.
Last edited by Tyler9000 on Wed Aug 03, 2016 10:50 am, edited 20 times in total.

bigred77
Posts: 2028
Joined: Sat Jun 11, 2011 4:53 pm

Re: A costly mistake made by both investors and advisors

Post by bigred77 » Tue Aug 02, 2016 5:50 pm

Da5id wrote:
HomerJ wrote:
larryswedroe wrote:But the math is what matters. It simply assumes valuations don't change because we cannot forecast if they will change.
Horrible assumption with no basis in reality that destroys the entire model.
What assumption would you make about earnings going forward if not the same? Do you think earnings are about to take off? Why are his assumptions horrible?

I don't know, sometimes it seems like the 4% rule is a religious belief, and to challenge it is heresy. The question is, apparently, whether Larry is heavier than a duck...
I think HomerJ dislikes that the model does not include changes in the P/E ratio as a variable. It makes sense why it's not used: because we don't have a reliable way to forecast it. There are some models that do try to incorporate it, but they don't do it very well.

I think Larry is incorporating a measure of future earnings growth into his model. That's a relatively well accepted practice I believe. We don't know what the future will hold but I'm fairly certain there will periods of significant earnings growth at some point (and also the reverse).

In the abstract, I agree with what Larry wrote. Valuations provide us with useful information. Over the next 10-20 years, real returns for US equities are more likely to under perform historical returns. Seems pretty clear cut and I agree with him. I am less certain about how useful valuations are in providing us information on the likelihood of US equities under performing their historical returns over 30-40 years (a time frame frequently used in retirement planning). I agree with HomerJ in that the Trinity study derived the 4% rule in order to be safe precisely when terrible returns rear their ugly head exactly during our planned retirement. Current valuations would not dissuade me from using the 4% in my retirement plan right now. I was already implementing a conservative plan, I do not see the need to be even more conservative.

lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 11:55 pm

Re: A costly mistake made by both investors and advisors

Post by lack_ey » Tue Aug 02, 2016 6:10 pm

HomerJ wrote:
larryswedroe wrote:But the math is what matters. It simply assumes valuations don't change because we cannot forecast if they will change.
Horrible assumption with no basis in reality that destroys the entire model.
It does not. It does suggest that there's a lot the model doesn't know about and thus the uncertainty is fairly high, as changes in valuations can make big differences in returns, especially in the shorter term. If you had a model that could predict where valuations are going to go with any kind of accuracy, it would improve your estimate. But many have tried and failed, and unless you have a good suggestion then this is just a starting place you can pick. It's kind of like assuming historical returns as a starting point for estimating future returns.

For what it's worth, some like Research Affiliates assume that valuations will on average revert to some kind of past mean and build models that incorporate that. Many of the earlier CAPE-based return projections you dislike so much implicitly or explicitly did this and were thus really really wrong about returns post-1995 or so.

Again, this is not saying that valuations will stay the same. They're pretty much guaranteed not to. Just that it's what we expect on average (again, the estimate is about the average return, where half the time the results should turn out above the average and half the time under the average, loosely speaking). Might be higher, might be lower. If it were more likely to be higher than lower or vice versa, we could incorporate that in the model.
HomerJ wrote:He said the expected return based on valuations is UNCONDITIONAL.

4.1% expected return over the next 1 year
4.1% expected return over the next 3 years
4.1% expected return over the next 10 years
4.1% expected return over the next 20 years
4.1% expected return over the next 30 years
4.1% expected return over the next 50 years

We ask "How can all these be accurate?"

The answer, "Oh, we assume valuations won't change"

But next year, they will have new "expected" returns to publish. Unconditionally. So useful.
I don't get why you're upset by this. That's how these things work. Of course this stuff isn't all that accurate, and it's not too controversial that new information can improve forecasts.

If in the next five years we have an annualized return of -6% a year, then the best 5-year forecast available at that time in 2021 is probably going to look different from either 4.1% or 15.3% (five years of 15.3% would result in a 10-year annualized return of 4.1% from 2016-2026), reflecting the new reality. That doesn't in of itself invalidate any of the current projections.


Now, I think it's very possible that some of the above assumptions are not particularly the best that could be made. But perhaps reasonable. And of course all these reflect a high degree of uncertainty regardless.

User avatar
siamond
Posts: 5125
Joined: Mon May 28, 2012 5:50 am

Re: A costly mistake made by both investors and advisors

Post by siamond » Tue Aug 02, 2016 8:43 pm

larryswedroe wrote:
siamond wrote:It would have been very useful to clarify that expected returns forecasts (with all the uncertainty coming with it) are about the coming 10 to 15 years, and NOT about the rest of our lives.
No, this is incorrect. When financial economists forecast returns they are UNCONDITIONAL returns, meaning regardless of the term.
Hm. I was trying to find an interpretation of your article that made sense, but it appears that I failed... I have no clue which financial economists you are referring to (references would be useful). I'll just point out that, when discussing expected returns, John Bogle makes a clear distinction between speculative returns and investment/enterprise returns, and sure enough, the former is a mid-term horizon and driven by speculative metrics using valuations, while the latter is for longer horizons for which averaged historical data is more relevant. Which only makes sense. I can understand your article in the first context (mid-term horizon), but definitely not in the second context (long-term horizon).
HomerJ wrote:
larryswedroe wrote:But the math is what matters. It simply assumes valuations don't change because we cannot forecast if they will change.
Horrible assumption with no basis in reality that destroys the entire model.
Indeed. This doesn't match whatsoever to any historical record. And totally ignores 'return-to-the-mean' past observations... :shock:

I have to say that I find it worrying to see a well-regarded and regular contributor to this forum making such assertions that the currently dire expected returns for both stocks and bonds are there to stay for the rest of our lives, and that the lessons of past history should simply be ignored. This is not only ludicrous, but worse, it is *very* misleading...
Last edited by siamond on Tue Aug 02, 2016 8:47 pm, edited 2 times in total.

2015
Posts: 2906
Joined: Mon Feb 10, 2014 2:32 pm

Re: A costly mistake made by both investors and advisors

Post by 2015 » Tue Aug 02, 2016 8:44 pm

Many thanks to Larry for this post, for his patience in explaining his logic, and to all who participated. These types of discussions on this forum are always very valuable in helping to clarify my thinking. I very much appreciate it!

Post Reply