My 30-year Forecast

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Rick Ferri
Posts: 8615
Joined: Mon Feb 26, 2007 11:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

My 30-year Forecast

Post by Rick Ferri » Thu Jan 09, 2014 12:23 pm

My 30-year market forecast for 2014 isn't much different from my 2013 forecast. I bumped US returns down a little and left developed market return as the were. For ROUGH planning purposes only!

Rick Ferri's 30-year Market Forecast for Investment Planning.

Please hold off judgement for 30-years if I'm still alive.

Rick Ferri
The Education of an Index Investor: born in darkness, finds indexing enlightenment, overcomplicates everything, embraces simplicity.

User avatar
Doc
Posts: 8743
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: My 30-year Forecast

Post by Doc » Thu Jan 09, 2014 12:33 pm

Thanks Rick,

I've had the page bookmarked now for several years. Glad to see you are still updating it.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

vencat
Posts: 214
Joined: Thu Sep 10, 2009 6:30 pm

Re: My 30-year Forecast

Post by vencat » Thu Jan 09, 2014 1:17 pm

Any reason for 30 years vs other time periods? Incidentally does anyone know why GMO does 7 year outlook every year???

Venkat

red5
Posts: 779
Joined: Sun Apr 01, 2012 4:42 pm

Re: My 30-year Forecast

Post by red5 » Thu Jan 09, 2014 1:23 pm

I for one am grateful to Mr. Rick Ferri for publishing these. I use the estimates in order to project a ***very rough estimate*** of how my numbers ***may*** look in the future. It gives me a vague sense of how I am doing. Thank you.

User avatar
Rick Ferri
Posts: 8615
Joined: Mon Feb 26, 2007 11:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

Re: My 30-year Forecast

Post by Rick Ferri » Thu Jan 09, 2014 1:35 pm

vencat wrote:Any reason for 30 years vs other time periods? Incidentally does anyone know why GMO does 7 year outlook every year???

Venkat
30 years tends to capture secular market returns and risks over a generation. Plus, I build portfolios looking out that far.

Rick Ferri
The Education of an Index Investor: born in darkness, finds indexing enlightenment, overcomplicates everything, embraces simplicity.

vencat
Posts: 214
Joined: Thu Sep 10, 2009 6:30 pm

Re: My 30-year Forecast

Post by vencat » Thu Jan 09, 2014 3:46 pm

Fair enough. Thanks.

User avatar
EternalOptimist
Posts: 829
Joined: Wed Nov 07, 2012 12:21 pm
Location: New York

Re: My 30-year Forecast

Post by EternalOptimist » Thu Jan 09, 2014 5:03 pm

Thanks Rick......lots of single digits.
"When nothing goes right....go left"

selftalk
Posts: 1096
Joined: Thu Mar 08, 2012 10:08 am

Re: My 30-year Forecast

Post by selftalk » Thu Jan 09, 2014 5:33 pm

Why did you lower developed small international companies and emerging markets ( which are in good corrections now thus higher returns in the future ) and then omit the frontier markets? What is your reasoning? Perhaps you may want to follow dr. Joel fuhrman as this may increase your chances of being here for 30 more years. Note his GBOMBS.

pop77
Posts: 370
Joined: Wed Dec 12, 2007 7:57 pm

Re: My 30-year Forecast

Post by pop77 » Thu Jan 09, 2014 5:49 pm

Great reference page. Helps me for my 'Value Averaging' projections. Two quick questions

1. All the NON-US Bonds are grouped under one category. Do you think this will be different for NON_US Developed Bonds Vs Emerging Market Bonds? Surprised to see a relatively low number here.
2. Do you think NON-US REITs (VNQI) would be similar to that of US REIT(VNQ)? I was encouraged by your previous post on diversification benefits of REITS. I own 3% of VNQ and 3% of VNQI in my portfolio.

User avatar
galeno
Posts: 1423
Joined: Fri Dec 21, 2007 12:06 pm

Re: My 30-year Forecast

Post by galeno » Thu Jan 09, 2014 5:57 pm

My calculated expected returns are almost identical to Ricks' except for bonds. On rebalance our FI's duration = 4.8 yr with a net nominal yield = 1.60%. That's a negative net real yield over the next 30 years for 40-60% of our retirement port.

If interest rates stay the same over the next 4 years the duration of our FI = 5.6 yr with a net nominal yield = 2.05%. Better. But probably still under water.

40-60% equities and a 3.0% AWR will have to finance our 39 year retirement.
AA = 40/55/5. Expected CAGR = 3.8%. GSD (5y) = 6.2%. USD inflation (10 y) = 1.8%. AWR = 4.0%. TER = 0.4%. Port Yield = 2.82%. Term = 33 yr. FI Duration = 6.0 yr. Portfolio survival probability = 95%.

snowman
Posts: 745
Joined: Thu Jan 31, 2013 12:59 pm

Re: My 30-year Forecast

Post by snowman » Thu Jan 09, 2014 6:32 pm

Thanks, Rick. I look forward to your forecast every year. Apart from bonds, where I expect 0% real return, I tend to agree with your rough estimates. Thanks again.

User avatar
Rick Ferri
Posts: 8615
Joined: Mon Feb 26, 2007 11:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

Re: My 30-year Forecast

Post by Rick Ferri » Thu Jan 09, 2014 6:48 pm

selftalk wrote:Why did you lower developed small international companies and emerging markets ( which are in good corrections now thus higher returns in the future ) and then omit the frontier markets?
You are correct that I did not include every segment of every market. Just the big ones.

Rick Ferri
The Education of an Index Investor: born in darkness, finds indexing enlightenment, overcomplicates everything, embraces simplicity.

garlandwhizzer
Posts: 2063
Joined: Fri Aug 06, 2010 3:42 pm

Re: My 30-year Forecast

Post by garlandwhizzer » Fri Jan 10, 2014 11:26 am

Thanks for this post, Rick. Very interesting, much to think about. I do have a question. You are expecting both REITS and US large cap stocks to give the same long term real returns, 5%. Have REITS historically performed as well as US large caps? Or are we entering a different era now?

Garland Whizzer

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: My 30-year Forecast

Post by Rodc » Fri Jan 10, 2014 12:17 pm

The long-term inflation expectations are 1.75%, according to The Federal Reserve Bank of Cleveland.
Three significant digits, I love it! :)

Predictions without error bars are entirely not-useful. Is an expectation of 2% meaningfully different from an expectation of 4%? Yes if error bars are +/- 1% and no if the error bars are +/- 10%.

With all due respect to Rick who is a very nice knowledgeable guy (if I pass and my wife engaged his services I would be entirely happy, well, if I knew), if anyone else published this we'd advise people not to pay attention to the noise. This should be treated as nothing more than playing around for fun, not something useful.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

User avatar
Garco
Posts: 707
Joined: Wed Jan 23, 2013 2:04 am

Re: My 30-year Forecast

Post by Garco » Fri Jan 10, 2014 1:52 pm

@Rodc, I respectfully disagree. It's not just for fun. We all have to make fundamental decisions about asset allocation, based on expected long-run returns. So it's valuable to have a working baseline for planning purposes. Otherwise, how would you set any kind of AA across different sectors? TSM, you say? Well how much should be allocated to international developed and developing markets? This does depend on expected returns and risk, relative to your own ability to accept risk or absorb loss over some time frame.

User avatar
arjking
Posts: 144
Joined: Mon Sep 03, 2007 9:33 am
Location: San Francisco

Re: My 30-year Forecast

Post by arjking » Fri Jan 10, 2014 2:25 pm

Rodc wrote: Predictions without error bars are entirely not-useful. Is an expectation of 2% meaningfully different from an expectation of 4%? Yes if error bars are +/- 1% and no if the error bars are +/- 10%.
I am interested to know the methodology. What kind statistical model was used?

Levett
Posts: 4177
Joined: Fri Feb 23, 2007 2:10 pm
Location: upper Midwest

Re: My 30-year Forecast

Post by Levett » Fri Jan 10, 2014 3:18 pm

I do appreciate 30-year forecasts, ancient as I am.

I will instruct my lawyer to attach said forecast as an addendum to my beneficiaries.

Vanguard Target 2055 sounds about right. ;-)

Lev

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: My 30-year Forecast

Post by Rodc » Fri Jan 10, 2014 4:25 pm

Garco wrote:@Rodc, I respectfully disagree. It's not just for fun. We all have to make fundamental decisions about asset allocation, based on expected long-run returns. So it's valuable to have a working baseline for planning purposes. Otherwise, how would you set any kind of AA across different sectors? TSM, you say? Well how much should be allocated to international developed and developing markets? This does depend on expected returns and risk, relative to your own ability to accept risk or absorb loss over some time frame.
You can't make meaningful decisions based on meaningless numbers. It might make you feel like you did something useful, but you didn't. Study after study has shown these sorts of market estimates have right about zero accuracy or information. No one gets these things right on a regular basis, though due to random luck once in a while someone guesses right.

The sad truth is asset allocation is an extremely crude business no matter how one dresses it up.
Last edited by Rodc on Fri Jan 10, 2014 4:30 pm, edited 1 time in total.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: My 30-year Forecast

Post by Rodc » Fri Jan 10, 2014 4:28 pm

arjking wrote:
Rodc wrote: Predictions without error bars are entirely not-useful. Is an expectation of 2% meaningfully different from an expectation of 4%? Yes if error bars are +/- 1% and no if the error bars are +/- 10%.
I am interested to know the methodology. What kind statistical model was used?
:confused

It is two hypothetical examples, one where there is useful information and one where there is not. Given Rick provided no error estimates at all, I have no idea which hypothetical might best match, and no methodology was use to produce error bars.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

Dulocracy
Posts: 2296
Joined: Wed Feb 27, 2013 1:03 pm
Location: Atlanta, GA

Re: My 30-year Forecast

Post by Dulocracy » Fri Jan 10, 2014 4:32 pm

I must admit that some of the numbers surprised me. While I understand this is more of an educated guess, I appreciate the time and effort that you put into it. It helps to have a realistic understanding of what to expect for planning purposes.
I'm not a financial professional. Post is info only & not legal advice. No attorney-client relationship exists with reader. Scrutinize my ideas as if you spoke with a guy at a bar. I may be wrong.

User avatar
Leif
Posts: 2499
Joined: Wed Sep 19, 2007 4:15 pm

Re: My 30-year Forecast

Post by Leif » Fri Jan 10, 2014 5:08 pm

Do you have an estimate for short term (2-3 year) bonds (gov. + corp)? TBM?

User avatar
Rick Ferri
Posts: 8615
Joined: Mon Feb 26, 2007 11:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

Re: My 30-year Forecast

Post by Rick Ferri » Sat Jan 11, 2014 9:10 am

Leif wrote:Do you have an estimate for short term (2-3 year) bonds (gov. + corp)? TBM?
My guess would be the expected inflation rate, 1.75%.

Rick Ferri
The Education of an Index Investor: born in darkness, finds indexing enlightenment, overcomplicates everything, embraces simplicity.

User avatar
youngindexer
Posts: 70
Joined: Wed Nov 20, 2013 5:55 pm

Re: My 30-year Forecast

Post by youngindexer » Sat Jan 11, 2014 10:00 am

Thanks for the forecast Rick!

givewell
Posts: 147
Joined: Sat Jun 11, 2011 5:39 pm

Re: My 30-year Forecast

Post by givewell » Sat Jan 11, 2014 10:43 am

Turning 72 in a few months, my 30 year forecast is that I won't be around 30 years from now. That is not certain, but it is highly likely.

About most things discussed on Bogleheads forums, nothing proposed is anywhere near that likely.

It is more nearly accurate to say that, on those topics, with 30 years as the time frame, nobody knows nuttin.

steve r
Posts: 506
Joined: Mon Feb 13, 2012 8:34 pm

Re: My 30-year Forecast

Post by steve r » Sat Jan 11, 2014 3:12 pm

Rick ... we can discuss this forecast at a Bogleheads 2044 meeting -- are you going to the one in Maui?
Maximize Diversification - Minimize Costs - Avoid Lotteries

abrahampayton
Posts: 24
Joined: Thu Oct 18, 2012 8:33 am
Location: Washington, D.C.
Contact:

Re: My 30-year Forecast

Post by abrahampayton » Sat Jan 11, 2014 5:00 pm

Thanks for posting, Mr. Ferri! I was looking at your forecast, and couldn't help but notice you mentioned "10-year tax-free municipal (A rated)" bonds.

So, I started snooping through Vanguard's muni bond fund selection and I noticed that the only muni bond fund they offered that had a maturity even close to 10 years was the High-Yield Tax-Exempt Fund (either VWAHX or VWALX). I've been looking at other Muni funds too (low-cost mutual funds or etfs). Anybody else find funds with this long of a duration? Seems like Vanguard's high-yield fund is the best bet.

User avatar
abuss368
Posts: 13017
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: My 30-year Forecast

Post by abuss368 » Sat Jan 11, 2014 5:04 pm

Hi Rick,

Thank you for the market forecast. I enjoy reading this every year and appreciate your perspective.

I found it interesting that you lowered your expected returns for the small value asset classes. Is it reasonable to assume this may be a result of more investors allocating their portfolio's to these asset classes due to popularity?

Best.
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

User avatar
Robert The Bruce
Posts: 81
Joined: Fri Mar 16, 2007 7:49 pm

Re: My 30-year Forecast

Post by Robert The Bruce » Sat Jan 11, 2014 7:35 pm

How can you adjust the 30 year return for half the globe from 1220% down to 760% with a simple "Emerging markets remain high at a 7.0 percent real return expectation, although that's lower than prevision forecasts due to slower growth in emerging market countries"? That's a step change, a huge step change.

I'm reminded of the "your cat is on the roof" joke - http://topjokes.blogspot.com/2006/10/ca ... -roof.html
The stone age didn't end for lack of stone.

User avatar
abuss368
Posts: 13017
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: My 30-year Forecast

Post by abuss368 » Sat Jan 11, 2014 8:24 pm

Has the 30 year forecast made anyone consider changing their allocation strategy to international funds? Considering increasing the international equity allocation?

We presently allocate 40% of equity to international.
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

555
Posts: 4955
Joined: Thu Dec 24, 2009 7:21 am

Re: My 30-year Forecast

Post by 555 » Sat Jan 11, 2014 10:39 pm

Rodc, what do mean no error bars? What do you think the "Risk Estimate" column gives?

User avatar
hoppy08520
Posts: 2023
Joined: Sat Feb 18, 2012 11:36 am

Re: My 30-year Forecast

Post by hoppy08520 » Sat Jan 11, 2014 11:54 pm

abuss368 wrote:I found it interesting that you lowered your expected returns for the small value asset classes. Is it reasonable to assume this may be a result of more investors allocating their portfolio's to these asset classes due to popularity?
That's exactly what his post on his website said:
Rick Ferri wrote:I’ve also lowered expectations for small-cap and value stocks. The popularity of these styles has attracted large cash flows in recent years and that tells me future expected returns may be lower.

User avatar
abuss368
Posts: 13017
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: My 30-year Forecast

Post by abuss368 » Sun Jan 12, 2014 12:45 am

hoppy08520 wrote:
abuss368 wrote:I found it interesting that you lowered your expected returns for the small value asset classes. Is it reasonable to assume this may be a result of more investors allocating their portfolio's to these asset classes due to popularity?
That's exactly what his post on his website said:
Rick Ferri wrote:I’ve also lowered expectations for small-cap and value stocks. The popularity of these styles has attracted large cash flows in recent years and that tells me future expected returns may be lower.
It is all good!
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

MBAkid24
Posts: 57
Joined: Fri Feb 15, 2013 2:53 pm

Re: My 30-year Forecast

Post by MBAkid24 » Sun Jan 12, 2014 5:37 am

Incidentally does anyone know why GMO does 7 year outlook every year???
We did a case in my MBA Finance class on GMO last month, and I had a classmate who worked for and knows Jeremy Grantham pretty well. They certainly aren't Bogleheads and believe that you can make macro-level calls about AA. Their whole platform is based on mean reversion within asset classes.

The 7 year time horizon is chosen because they believe it's enough time for a reasonable level of mean-reversion to occur. For example, if stocks are looking bubbly in their eyes, they don't think they can do a good 3 year forecast, because maybe the bubble will keep inflating. They also don't think 15 years is reasonable, because maybe another bubble has formed, and regardless all of their clients would have moved on. 7 is long enough for markets to correct (in their mind), but short enough to be within the "realm of patience" for a typical long-term client.

It's worth noting that they have more people engaged in marketing their expertise than they have studying the markets. If you think strategically about why he publishes his Quarterly outlook for free, it becomes clear that it is all brand-building / marketing expense. Not to say he's anything but brilliant... but it's worth thinking about.

Personally, I'll stick to Target Retirement 2040, thank you very much.

User avatar
jeffyscott
Posts: 7368
Joined: Tue Feb 27, 2007 9:12 am
Location: Wisconsin

Re: My 30-year Forecast

Post by jeffyscott » Sun Jan 12, 2014 8:08 am

vencat wrote:Any reason for 30 years vs other time periods? Incidentally does anyone know why GMO does 7 year outlook every year???

Venkat
For the same reason Rick does a 30 year forecast every year, things change.

GMO updates their forecast monthly, actually. The forecast is based on the market reaching fair value at the 7 year point. The estimates of fair value typically do not change much. But if fair value is, say 2000, the 7 year return to get there is going to be different if the market is already at 1800 than if it is at 1300.
press on, regardless - John C. Bogle

kv968
Posts: 6
Joined: Sun Jan 05, 2014 8:31 am

Re: My 30-year Forecast

Post by kv968 » Sun Jan 12, 2014 8:11 am

Rodc wrote:You can't make meaningful decisions based on meaningless numbers. It might make you feel like you did something useful, but you didn't. Study after study has shown these sorts of market estimates have right about zero accuracy or information. No one gets these things right on a regular basis, though due to random luck once in a while someone guesses right.

The sad truth is asset allocation is an extremely crude business no matter how one dresses it up.

...if anyone else published this we'd advise people not to pay attention to the noise. This should be treated as nothing more than playing around for fun, not something useful.
I kinda have to agree with Rodc. I respect Mr. Ferri and the work he's done, have read some of his books and think some of his estimates sound right "for now".

I realize you need some sort of jumping off point in order to develop an AA that will suit your needs but I also find it somewhat funny that investors who are totally against even "overweighting" or "underweighting" a sector or even a major asset category because of recent performance can reasonably think that a 30-year prediction can even be close to accurate. There also is some recency bias in some of the predictions in that Mr. Ferri feels that the 30-yr performance of small caps and value will be lower just because they've done well in the past few years..."I’ve also lowered expectations for small-cap and value stocks. The popularity of these styles has attracted large cash flows in recent years and that tells me future expected returns may be lower." What does that have to do 30 years from now?

And as Rodc also said, be honest, if anyone else published this they would be bashed with "predicting the future" and not to be taken seriously since no one knows (which they don't) what the future holds and they're just spouting "noise". Again, I respect Mr. Ferri's well-thought out estimations, understand there is a need for such things and realize they're only "expected" returns, but come on...30 years ago I was jamming that great new tune "Jump" from Van Halen on my boombox as I went to the mall to catch "A Nightmare On Elm Street", buying some beenie babies on my way out so my retirement future was secure, hoping I could make it home in time to catch that new "Miami Vice" show everyone was talking about :D

555
Posts: 4955
Joined: Thu Dec 24, 2009 7:21 am

Re: My 30-year Forecast

Post by 555 » Sun Jan 12, 2014 1:07 pm

Rodc, kv968, why are you pretending that the "Risk Estimate" column is not there when it is?

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: My 30-year Forecast

Post by Rodc » Sun Jan 12, 2014 1:44 pm

555 wrote:Rodc, kv968, why are you pretending that the "Risk Estimate" column is not there when it is?
He does not provide an estimate of the error in his 30 year forecast if you read more carefully what that column is. He simply says it will vary year to year about his estimate by some amount. That is an entirely different issue.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

555
Posts: 4955
Joined: Thu Dec 24, 2009 7:21 am

Re: My 30-year Forecast

Post by 555 » Sun Jan 12, 2014 1:46 pm

Rodc wrote:
555 wrote:Rodc, kv968, why are you pretending that the "Risk Estimate" column is not there when it is?
He does not provide an estimate of the error in his 30 year forecast if you read more carefully what that column is. He simply says it will vary year to year about his estimate by some amount. That is an entirely different issue.
Let me ask a different way. What exactly is it that is being claimed to be being "forecast"?

boggler
Posts: 595
Joined: Thu Feb 07, 2013 2:29 am

Re: My 30-year Forecast

Post by boggler » Sun Jan 12, 2014 2:01 pm

Doesn't the efficient market hypothesis imply that all asset classes and components will be weighted in such a way as to make the expected return for each one proportional to its risk? If so, how can we even predict that US will do better than international, or whatever? Shouldn't our predictions just be the same for everything?

Leeraar
Posts: 4109
Joined: Tue Dec 10, 2013 8:41 pm
Location: Nowhere

Re: My 30-year Forecast

Post by Leeraar » Sun Jan 12, 2014 2:16 pm

Yes, Mr. Ferri said "Forecast", maybe he should have said "Outlook".

I am quite comfortable with this, and think it is entirely rational to pontificate on "expected" values. The usual answer is, we'll muddle along, and things will be much as they are now. Or, things that are out of whack are likely to revert to historical values.

Here is my personal outlook: What is the outlook for interest rates? Low now, likely to remain low in the near future but to rise in the longer term. What about inflation? Well, there are powerful regulatory forces that want to see inflation at about 2%. Don't change your strategy because someone posted the picture of the guy with a wheelbarrow full of banknotes.

There are lots of things that can be predicted quite closely in the aggregate, but the predictions are not very useful for taking specific actions. Examples include population demographics, traffic fatalities, homicide rates, deaths due to influenza, influenza vaccination rates, etc.

L.

(Edited for a typo.)
Last edited by Leeraar on Sun Jan 12, 2014 3:32 pm, edited 1 time in total.
You can get what you want, or you can just get old. (Billy Joel, "Vienna")

555
Posts: 4955
Joined: Thu Dec 24, 2009 7:21 am

Re: My 30-year Forecast

Post by 555 » Sun Jan 12, 2014 3:10 pm

Leeraar wrote:Yes, Mr. Ferri said "Forecast", maybe he should have said "Outlook".
No the word doesn't really matter. The problem is that some people are mischaracterizing the nature of the "Forecast".

The "Risk Estimate" column, expresses a large amount of uncertainty, and some people are conveniently assuming it's not there, to make some point.

User avatar
LadyGeek
Site Admin
Posts: 49203
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: My 30-year Forecast

Post by LadyGeek » Sun Jan 12, 2014 3:15 pm

Rick has graciously given permission to use his 30 year forecast in the wiki: Historical and expected returns

The "U.S. micro-cap stocks" entry is not in the 2013 version. Here's the 2012 version: Revision as of 20:02, 3 November 2013

=============================================
I should also note that the wiki update request was received by use of the Reader Feedback Form, which is at the bottom of every wiki page. No login required, just fill out the form.

We promote helpful comments to the Discussion tab, as shown here: Talk:Historical and expected returns
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: My 30-year Forecast

Post by Rodc » Sun Jan 12, 2014 3:42 pm

555 wrote:
Rodc wrote:
555 wrote:Rodc, kv968, why are you pretending that the "Risk Estimate" column is not there when it is?
He does not provide an estimate of the error in his 30 year forecast if you read more carefully what that column is. He simply says it will vary year to year about his estimate by some amount. That is an entirely different issue.
Let me ask a different way. What exactly is it that is being claimed to be being "forecast"?
Maybe an overview of some history and math would help.

A few hundred years ago Gauss pioneered least squares and it was refined by others like Markov. The original use was in predicting planetary orbits. Different people made measurements of different quality and the issue was how to best combine these measurement to get an optimal estimate. You need at a minimum two parameters per measurement, mean and variance (square of standard deviation). Then you also need the correlation. Two measurements that are fully correlated have no more information than one independent measurement and that has to be taken into account. So you need the pairwise correlation between all pairs of measurements.

Harry Markowitz borrowed from this mathematical approach and published his first paper on this idea in 1952, and subsequently refined the ideas and eventually in 1987 received the "Nobel" in Economic. This of course requires estimates of the mean and variance of each asset, and correlations. (Clearly you need correlations because you can't treat large cap US as being completely independent of small cap US).

So to start, Rick has not provided the information to be actionable - there are no correlations and correlations are extremely important in understanding how to use mean and variance which he did provide (or almost provided as variance is just the square of the provided standard deviation).

We see many threads here where people examine asset allocations using a variety of methods. I do not recall one person ever directly using mean and standard deviation. Why is that? Well you need correlation for one, but presumably one could estimate that. Indeed once in a while someone posts some data on correlations.

The reason is that soon after Markowitz developed his ideas, soon to be called Modern Portfolio theory, people figured out that it routinely gave clearly bogus results. The problem is that it is mathematically very sensitive to errors in the input parameters. Minor changes in inputs gave wildly different weights!

And no one can provide estimates of mean, variance and correlations that are accurate enough to tame this bad sensitivity.

So my complaint was a simplification of the entire problem. I focused on mean to simplify things. Really Rick has provided only two of the three required input, and has said nothing about how well he can estimate the two he did provide.

Think of a simple example - a coin toss game with a fair coin. Suppose you don't have a nice model of a coin and you have to estimate from data. Of course if this is an actual fair coin the odds of heads on any given toss are 50%. But suppose you toss the coin 10 times and see 6 heads. So you estimate the odds of a head is 60%. You also know that toss to toss there will be lots of variability on the way to getting about 60% heads. (This is like an estimate of 5% real for large US over 30 annual tosses with a standard deviation of 19%).

If you think it is 60% +/- 1% you might be willing to bet a lot on a bet that in 100 tosses more than 55 heads will happen. But if you think that it is 60% +/- 10% you might pass on a large bet of getting 55 or more. Tying this back to the original post, if asset one has an expected return of 5% and standard deviation of 19% and some other asset has an expected return of 7% and a standard deviation of 19% (so more return and same risk) I would load up to some degree on the second if I really felt those estimates of returns were very accurate. On the other hand if I thought those estimates where very inaccurate I would tend to get about half each since I would have little confidence in picking one over the other.

One more point might be worth mentioning.

Population mean vs sample mean. The population mean for a fair coin toss (if you flip a coin an infinite number of times) is 50% heads and 50% tails. In a sample of 10 tosses the sample mean might as above be 60/40.

The reason this is important is that for the stock market you never know the population mean; you can only know the sample mean (standard deviation, correlation, etc). It is actually worse than that, but the bottom line is we don't even know the past population mean to better than +/- 2% or so, because all we have is a sample mean, so - how can estimates of the future population mean be more accurate than that? (note strategy must be based on population stats or "true" stats, not sample stats which are full of noise. Your actually results of course depend on what sample you happen to draw, unfortunately.)

Anyone who thinks that they can use estimates of mean, standard deviation, (without correlation no less) to do meaningful portfolio construction is likely quite mistaken and especially so if they have no good idea of how accurate their mean and standard deviation estimates are. IMHO

Typos and some additions made 4:22 pm East Coast time
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

stlutz
Posts: 4839
Joined: Fri Jan 02, 2009 1:08 am

Re: My 30-year Forecast

Post by stlutz » Sun Jan 12, 2014 4:29 pm

I'm not quite sure what to make of some of the back-and-forth here. A few of my thoughts:

First, on the use of the word "forecast", I've always read this with a hint of irony--this time of year everyone in the media is offering up their 2014 forecasts; well Rick is offering his forecast for the next 30 years! That said, I do appreciate and support his choice of words. It's a refreshing contrast to those who will come on here and say something like, "THE expected return for bonds is 1%, for large stocks is 2%, for small-value stocks is 9%, emerging markets-small-value stocks is 15%, and Canadian midcap value Timber REITS is 20%." These types of projections don't come down from on high. Rather, investors make them based on their understandings of how markets work and the world we live in. I appreciate the fact that Rick says these the *his* forecasts, not the one correct "expected" return calculation.

Second, someone's forecasts can change because the markets changed or because their own views changed. My perception having been around here a while is that Rick has become somewhat less enthusiastic about various forms of tilting over the past few years. He still believes that small/value and emerging will "win", just not by as much. (I'm sure Rick will correct me if my perception is wrong!).

Smallcap value stocks represent only a few percent of the overall market. The overall market doesn't have to change it's opinion of such stocks by that much for all future "expected" outperformance to be wiped out. With the popularity of all sorts of quantitative value strategies ("spindexing"), caution in this area is warranted in my view.

On emerging markets, I don't see why one should expect any long-term return premium there. The transition from state-run to partially-market-driven economies has already occurred in these places. There is perhaps more potential in "frontier" markets, but these of course also have a lot of challenges with basic political stability.

Finally, asset allocation is always an approximate affair. Having error bars in such forecasts suggests a level of mathematical rigor to these types of "forecasts" that simply doesn't exist. Everybody has expectations when they invest; some folks think about these things in more numerical ways; others are more qualitative ("I think stocks will beat bonds by a pretty good margin, and small value might eek out a little more"). I don't know that anybody here expects anyone's forecasts to be "spot on." If I'm wrong and anyone does have this expectation, please speak up. :happy

richard
Posts: 7961
Joined: Tue Feb 20, 2007 3:38 pm
Contact:

Re: My 30-year Forecast

Post by richard » Sun Jan 12, 2014 4:35 pm

Rodc wrote:The reason this is important is that for the stock market you never know the population mean; you can only know the sample mean (standard deviation, correlation, etc). It is actually worse than that, but the bottom line is we don't even know the past population mean to better than +/- 2% or so, because all we have is a sample mean, so - how can estimates of the future population mean be more accurate than that? (note strategy must be based on population stats or "true" stats, not sample stats which are full of noise. Your actually results of course depend on what sample you happen to draw, unfortunately.)
We don't even know there is a stable population mean, let alone know what it is.

How did you derive +/- 2%?

People hunger for certainty in a chaotic world. Combining that with a lack of a good understanding of statistics may explain the enthusiasm for market forecasts. OTOH, how does one determine a savings rate (let alone an asset allocation) without some sort of estimate of future returns?

richard
Posts: 7961
Joined: Tue Feb 20, 2007 3:38 pm
Contact:

Re: My 30-year Forecast

Post by richard » Sun Jan 12, 2014 4:39 pm

stlutz wrote:Finally, asset allocation is always an approximate affair. Having error bars in such forecasts suggests a level of mathematical rigor to these types of "forecasts" that simply doesn't exist. Everybody has expectations when they invest; some folks think about these things in more numerical ways; others are more qualitative ("I think stocks will beat bonds by a pretty good margin, and small value might eek out a little more"). I don't know that anybody here expects anyone's forecasts to be "spot on." If I'm wrong and anyone does have this expectation, please speak up. :happy
Forecasting 30 year returns to one decimal place (let alone more) suggests a level of mathematical rigor to these types of "forecasts" that simply doesn't exist.

Which do you believe better communicates the inherent uncertainty: predicting 30 year returns at 2%-6% or predicting 4.2%?

BTW, forecasts that purport to be made to a higher degree of accuracy, or generally that are made with more confidence, tend to be more believed but less accurate. See Philip Tetlock's Expert Political Judgement.

stlutz
Posts: 4839
Joined: Fri Jan 02, 2009 1:08 am

Re: My 30-year Forecast

Post by stlutz » Sun Jan 12, 2014 4:47 pm

Forecasting 30 year returns to one decimal place (let alone more) suggests a level of mathematical rigor to these types of "forecasts" that simply doesn't exist.

Which do you believe better communicates the inherent uncertainty: predicting 30 year returns at 2%-6% or predicting 4.2%?
I'm more of a qualitative thinker, so when I see large cap stocks at 5% and smallcap stocks at 5.3%, I read this as "Small stocks will probably beat large stocks buy just a little bit." Thinking numerically, I'm not sure how you express this sentiment without the use of a decimal point.

Again, if anyone is really thinking that the future will be exactly as Rick has outlined, they should speak up. Otherwise we're setting up a straw man, making a mountain out of a molehill, or beating a dead horse here.

richard
Posts: 7961
Joined: Tue Feb 20, 2007 3:38 pm
Contact:

Re: My 30-year Forecast

Post by richard » Sun Jan 12, 2014 4:55 pm

stlutz wrote:
Forecasting 30 year returns to one decimal place (let alone more) suggests a level of mathematical rigor to these types of "forecasts" that simply doesn't exist.

Which do you believe better communicates the inherent uncertainty: predicting 30 year returns at 2%-6% or predicting 4.2%?
I'm more of a qualitative thinker, so when I see large cap stocks at 5% and smallcap stocks at 5.3%, I read this as "Small stocks will probably beat large stocks buy just a little bit." Thinking numerically, I'm not sure how you express this sentiment without the use of a decimal point.
What's wrong with "Small stocks will probably beat large stocks by just a little bit"?

5.3% implies an undue level of precision and, due to cognitive biases, tends to be more believed.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: My 30-year Forecast

Post by Rodc » Sun Jan 12, 2014 5:09 pm

richard wrote:
Rodc wrote:The reason this is important is that for the stock market you never know the population mean; you can only know the sample mean (standard deviation, correlation, etc). It is actually worse than that, but the bottom line is we don't even know the past population mean to better than +/- 2% or so, because all we have is a sample mean, so - how can estimates of the future population mean be more accurate than that? (note strategy must be based on population stats or "true" stats, not sample stats which are full of noise. Your actually results of course depend on what sample you happen to draw, unfortunately.)
We don't even know there is a stable population mean, let alone know what it is.

How did you derive +/- 2%?

People hunger for certainty in a chaotic world. Combining that with a lack of a good understanding of statistics may explain the enthusiasm for market forecasts. OTOH, how does one determine a savings rate (let alone an asset allocation) without some sort of estimate of future returns?
I agree. A back of the envelop calculation based on a super simple ideal world gives the 2%, hence the +/- 2% or so.

If the world were Gaussian (simplification 1), underlying forces never change (simplification 2), year to year independent (simplification 3), we had 100 years of totally accurate returns (simplification 4) and we measured a standard deviation of 20% (not far off from the historical record for stocks) the standard deviation of the error between population mean and sample mean would be estimated as 20%/sqrt(100) = 2%.

As noted that includes a string of idealizations. So in reality likely the number is bigger. But that strikes me as sufficient to illustrate the difficulties. My sense is most people have no idea of even a rough magnitude for this error. It makes it clear the difference between 5.0% and 5.3% is pure noise.

Added: as Richard notes I should have listed as a first simplification that there is A population mean.

I often include all of these caveats when I make this point, but one of the challenges when posting as opposed to write a longer paper or a book is at some point it gets too long for many readers. This format is made for fairly short posts and you just can't include everything. You cut some things to make the main point. In this case I elected to not to get into the above, but Richard's point is certainly valid.
Last edited by Rodc on Sun Jan 12, 2014 5:22 pm, edited 3 times in total.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: My 30-year Forecast

Post by Rodc » Sun Jan 12, 2014 5:14 pm

stlutz wrote:
Forecasting 30 year returns to one decimal place (let alone more) suggests a level of mathematical rigor to these types of "forecasts" that simply doesn't exist.

Which do you believe better communicates the inherent uncertainty: predicting 30 year returns at 2%-6% or predicting 4.2%?
I'm more of a qualitative thinker, so when I see large cap stocks at 5% and smallcap stocks at 5.3%, I read this as "Small stocks will probably beat large stocks buy just a little bit." Thinking numerically, I'm not sure how you express this sentiment without the use of a decimal point.

Again, if anyone is really thinking that the future will be exactly as Rick has outlined, they should speak up. Otherwise we're setting up a straw man, making a mountain out of a molehill, or beating a dead horse here.
if it I think about 5% but really is in the range of 2% to 8% vs I think about 5.3% and a range of 1% to 10% would you come to the same conclusion?

How about 5% +/- 10% vs 5.3% +/- 10%?

In the later case the correct conclusion is they are tied as the 0.3 part is meaningless.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

Post Reply