William Bernstein: the paradox of wealth

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
grok87
Posts: 8295
Joined: Tue Feb 27, 2007 9:00 pm

William Bernstein: the paradox of wealth

Post by grok87 » Mon Oct 07, 2013 4:44 pm

...
http://www.cfainstitute.org/learning/pr ... .n5.1.aspx
..
Another great article by Dr. Bernstein. He makes the case that stock and bond returns will be more muted going forward. He thinks fair value for Shillers PE10 may be 20 now instead of the historical average of 16.
Bernstein wrote: The familiar 10-year cyclically adjusted price-to-earning ratio (CAPE) series of US stocks...suggests that over the past 132 years earnings muliples have expanded with an average annual regression slope of 0.058.[But] the t-statstic for the 14 independent 10 year CAPE points is only 1.65, close but no cigar.
Keep calm and Boglehead on. KCBO.

grok87
Posts: 8295
Joined: Tue Feb 27, 2007 9:00 pm

Re: William Bernstein: the paradox of wealth

Post by grok87 » Mon Oct 07, 2013 5:49 pm

Here's a better (i.e. more accessible) link
http://larrysiegeldotorg.files.wordpres ... matted.pdf
cheers,
Keep calm and Boglehead on. KCBO.

User avatar
jjustice
Posts: 470
Joined: Tue Aug 04, 2009 9:58 am

Re: William Bernstein: the paradox of wealth

Post by jjustice » Mon Oct 07, 2013 6:55 pm

By serendipity, the NY Times just today had an essay by Stephen King (not the novelist), "When Wealth Disappears" (http://www.nytimes.com/2013/10/07/opini ... .html?_r=0) making exactly the sort of argument that Bernstein criticizes. I am not capable of adjudicating the dispute, but I was quite pleased to see Bernstein make Karl Popper's point that technological breakthroughs cannot be known in advance, since if they could be, they would have already occurred.

Bernstein thinks that it is more reasonable to project the continuation of technological breakthroughs than to predict a continuing increase in coal production.

John

Bill Bernstein
Posts: 557
Joined: Sat Jun 23, 2007 12:47 am

Re: William Bernstein: the paradox of wealth

Post by Bill Bernstein » Mon Oct 07, 2013 7:25 pm

The version on Larry Siegel's website is a much expanded version of what actually appeared in the FAJ, which deep-sixed the part about overall economic growth. Only the part about lower security returns made it into print.

Bill

larryswedroe
Posts: 15691
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: William Bernstein: the paradox of wealth

Post by larryswedroe » Mon Oct 07, 2013 7:50 pm

FWIW IMO the piece by Bill is brilliant work and his conclusions are very consistent with all the literature I've read on economic growth and stock returns
Larry

MIretired
Posts: 733
Joined: Fri Sep 06, 2013 12:35 pm

Re: William Bernstein: the paradox of wealth

Post by MIretired » Mon Oct 07, 2013 9:20 pm

Great article. Easy read. Uncomplicated. I'll reread it. Inspiring.

User avatar
LazyNihilist
Posts: 874
Joined: Sat Feb 19, 2011 9:56 pm
Location: 6.66% (xirr)
Contact:

Re: William Bernstein: the paradox of wealth

Post by LazyNihilist » Mon Oct 07, 2013 10:27 pm

I think there is one instance of a society running out of "raw material". Easter Island. May be it's an anomaly.
It's crazy to think that after the fall of Rome it took Europe close to 1500 years to get back to that level of prosperity.
The strong do what they can and the weak suffer what they must -Thucydides

grok87
Posts: 8295
Joined: Tue Feb 27, 2007 9:00 pm

Re: William Bernstein: the paradox of wealth

Post by grok87 » Mon Oct 07, 2013 10:32 pm

jjustice wrote:By serendipity, the NY Times just today had an essay by Stephen King (not the novelist), "When Wealth Disappears" (http://www.nytimes.com/2013/10/07/opini ... .html?_r=0) making exactly the sort of argument that Bernstein criticizes. I am not capable of adjudicating the dispute, but I was quite pleased to see Bernstein make Karl Popper's point that technological breakthroughs cannot be known in advance, since if they could be, they would have already occurred.

Bernstein thinks that it is more reasonable to project the continuation of technological breakthroughs than to predict a continuing increase in coal production.

John
THanks. the last paragraph of the NY Times article is interesting...
Keep calm and Boglehead on. KCBO.

Angst
Posts: 1823
Joined: Sat Jun 09, 2007 11:31 am

Re: William Bernstein: the paradox of wealth

Post by Angst » Mon Oct 07, 2013 11:26 pm

Thank you grok for bringing this to our attention! And to jjustice for the link to the King article.

After reading Dr. Bernstein's interesting and entertaining review of the history and development of the current state of the world economy, and noting how he used this broad perspective to consider where our economy might be progressing to today, I found the gloom and doom of Mr. King's article in the NY Times almost comical in the ernest and fairly foregone nature of its conclusions. I much prefered Dr. Bernstein's analysis and way of thinking.

User avatar
shmidds
Posts: 152
Joined: Fri Jun 12, 2009 6:25 pm

Re: William Bernstein: the paradox of wealth

Post by shmidds » Tue Oct 08, 2013 1:03 am

Very timely...I'm halfway through "The Birth of Plenty." Laid down the book and then read this article, should have gone to bed four hours ago.

User avatar
3CT_Paddler
Posts: 3222
Joined: Wed Feb 04, 2009 5:28 pm
Location: Marietta, GA

Re: William Bernstein: the paradox of wealth

Post by 3CT_Paddler » Tue Oct 08, 2013 7:03 am

Interesting stuff. It certainly makes intuitive sense that as a group or nation becomes wealthier (has more capital to spend), the magnitude of the opportunities for that capital is gradually reduced. I think the Great Recession was a modern day short term example of that reality. With capital for investment relatively scarce, the opportunities for growth were much better in 2009 than they are today.

KyleAAA
Posts: 6667
Joined: Wed Jul 01, 2009 5:35 pm
Contact:

Re: William Bernstein: the paradox of wealth

Post by KyleAAA » Tue Oct 08, 2013 7:25 am

Though there are plenty of civilizations that collapsed because they ran out of resources, from an overall global perspective this analysis makes intuitive sense to me.

User avatar
HomerJ
Posts: 11564
Joined: Fri Jun 06, 2008 12:50 pm

Re: William Bernstein: the paradox of wealth

Post by HomerJ » Tue Oct 08, 2013 9:29 am

We are reaching end times for Western affluence. Between 2000 and 2007, ahead of the Great Recession, the United States economy grew at a meager average of about 2.4 percent a year — a full percentage point below the 3.4 percent average of the 1980s and 1990s. From 2007 to 2012, annual growth amounted to just 0.8 percent. In Europe, as is well known, the situation is even worse. Both sides of the North Atlantic have already succumbed to a Japan-style “lost decade.”
So the 18-year bull market of the 80s and 90s has been followed by 13 years of bear market...

And this equals end-times?

Valuethinker
Posts: 35624
Joined: Fri May 11, 2007 11:07 am

Re: William Bernstein: the paradox of wealth

Post by Valuethinker » Tue Oct 08, 2013 9:49 am

jjustice wrote:By serendipity, the NY Times just today had an essay by Stephen King (not the novelist), "When Wealth Disappears" (http://www.nytimes.com/2013/10/07/opini ... .html?_r=0) making exactly the sort of argument that Bernstein criticizes. I am not capable of adjudicating the dispute, but I was quite pleased to see Bernstein make Karl Popper's point that technological breakthroughs cannot be known in advance, since if they could be, they would have already occurred.

Bernstein thinks that it is more reasonable to project the continuation of technological breakthroughs than to predict a continuing increase in coal production.

John
As investors we should be aware there have been long periods (the late 19th century) when investments made poor returns, and when economies were stuck in stagnation. That's more or less what King (former Chief Economist HSBC bank?) is writing about.

The counterpoint on energy is Vaclav Smil, who shows you in several books just how slowly methods of energy production change in terms of total consumption. Coal was the world's largest energy source in 1930, and it still is number 2 in 2013 (might be marginally lapped by natural gas). Oil is something like 35-40% world energy consumption and has been since 1950 at least. Nuclear went from 0 to 8% but is now slipping. Natural gas is the big gainer over that time period and looks like the one that will make further gains in the 21st century (it's cleaner than coal, but it does not solve the problem). The oldest form of human energy, biomass, is still a big component, despite all the inherent environmental problems (smoke from cooking kills thousands if not millions of 3rd worlders very year).

Wind and solar are tiny--less than 2%. Despite exponential growth in installations. Most 'renewables' are in fact, hydro electric, almost the oldest source of electric power.

The message is one of a certain Long Run optimism, but also of Realism. Societies tend to access their cheapest resources first regardless of the long run harm. Even though this has at various times led to disastrous deforestation (the Romans and Greeks), soil erosion and even collapse.

What we are asking is humans to break an age old pattern of favouring the short term over the long term-- consider the state of fish stocks in our oceans. *that* is very difficult, it runs counter to human nature. There are plenty of examples in history where it was done (Elinor Ostrin won the Nobel Prize for her studies of communities managing common resources) but probably even more when it has not.

Hope is not a plan. 'Something will turn up' cannot be the basis for what you do.

Another way of putting it, sometimes called 'the Precautionary Principle' is 'Praise Allah, but first tie up your camel'.

thx1138
Posts: 774
Joined: Fri Jul 12, 2013 2:14 pm

Re: William Bernstein: the paradox of wealth

Post by thx1138 » Tue Oct 08, 2013 9:54 am

HomerJ wrote:
We are reaching end times for Western affluence. Between 2000 and 2007, ahead of the Great Recession, the United States economy grew at a meager average of about 2.4 percent a year — a full percentage point below the 3.4 percent average of the 1980s and 1990s. From 2007 to 2012, annual growth amounted to just 0.8 percent. In Europe, as is well known, the situation is even worse. Both sides of the North Atlantic have already succumbed to a Japan-style “lost decade.”
So the 18-year bull market of the 80s and 90s has been followed by 13 years of bear market...

And this equals end-times?
King doesn't seem to reference the stock market at all, he's talking about real GDP growth. And that is much lower now than it was in previous long bear markets (during which GDP growth was actually still quite high).

Of course whether that is predictive in any way is a whole different ball of wax. I'm going to go with "not predictive" myself. Or alternatively "beware recency bias".

The title of King's book looks like it would be a good choice to put in Bill's introduction if he republishes his paper in a decade or so :happy

User avatar
BlueEars
Posts: 3614
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Re: William Bernstein: the paradox of wealth

Post by BlueEars » Tue Oct 08, 2013 10:24 am

I noticed Bernstein is not quite convinced about that regression plot on PE10 (Figure 8):
As we’ve seen, it takes centuries for wealth to drive down security returns, and the data
from the Shiller series is not quite long enough to demonstrate convincingly the above
noted positive regression slope; its t-stat for the 14 independent 10-yearCAPE datapoints is just 1.65
—close, but no cigar yet; maybe in another century or two, we’ll know
for sure if equity returns have fallen in the same way that loan rates have.
So we are off the hook here. I can still hope for good equity returns! :happy

Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Re: William Bernstein: the paradox of wealth

Post by Browser » Tue Oct 08, 2013 11:05 am

I don't quite get what he's saying about PE10 - can someone translate?
The longest continuous series of high-quality equity valuation data, the familiar Robert
Shiller 10-year cyclically adjusted price/earnings ratio (CAPE) series of U.S. stocks,
suggests that over the past 132 years, earnings multiples have expanded, with a
regression slope of 0.058 points/per year. In other words, this widely followed ratio
seems to increase by a point every 17 years or so; the intercepts of this trendline are 13.6
in 1881 and 20.3 at year-end 2012, at which point the actual value was 21.3.

As we’ve seen, it takes centuries for wealth to drive down security returns, and the data
from the Shiller series is not quite long enough to demonstrate convincingly the above
noted positive regression slope; its t-stat for the 14 independent 10-year CAPE data
points is just 1.65—close, but no cigar yet; maybe in another century or two, we’ll know
for sure if equity returns have fallen in the same way that loan rates have.
In exactly what way are the PE10 data indicating the possibility that equity returns may be falling?
We don't know where we are, or where we're going -- but we're making good time.

User avatar
HomerJ
Posts: 11564
Joined: Fri Jun 06, 2008 12:50 pm

Re: William Bernstein: the paradox of wealth

Post by HomerJ » Tue Oct 08, 2013 11:20 am

thx1138 wrote:Of course whether that is predictive in any way is a whole different ball of wax. I'm going to go with "not predictive" myself. Or alternatively "beware recency bias".
Exactly... 13 years of poor numbers does not equal "end-times".

Seems like a ton of economists are falling prey to "recency basis"

User avatar
jjustice
Posts: 470
Joined: Tue Aug 04, 2009 9:58 am

Re: William Bernstein: the paradox of wealth

Post by jjustice » Tue Oct 08, 2013 11:46 am

Browser wrote:I don't quite get what he's saying about PE10 - can someone translate?
In exactly what way are the PE10 data indicating the possibility that equity returns may be falling?
PE10 is a measure of how many dollars of capital must be invested to get a dollar of return. So if it is taking more capital for the same return, it would be reflected in P/E ratios.

Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Re: William Bernstein: the paradox of wealth

Post by Browser » Tue Oct 08, 2013 12:01 pm

jjustice wrote:
Browser wrote:I don't quite get what he's saying about PE10 - can someone translate?
In exactly what way are the PE10 data indicating the possibility that equity returns may be falling?
PE10 is a measure of how many dollars of capital must be invested to get a dollar of return. So if it is taking more capital for the same return, it would be reflected in P/E ratios.
OK, now explain to me how PE10 is a measure of investment dollars per unit of return and I'll understand.
We don't know where we are, or where we're going -- but we're making good time.

thx1138
Posts: 774
Joined: Fri Jul 12, 2013 2:14 pm

Re: William Bernstein: the paradox of wealth

Post by thx1138 » Tue Oct 08, 2013 1:01 pm

Browser wrote: OK, now explain to me how PE10 is a measure of investment dollars per unit of return and I'll understand.
Well P is Price, and E is Earnings. Recent earnings are the best known ex ante indicator of future returns, though of course an imperfect indicator. And the dividend discount model estimates the current price as being the sum total of expected discounted future returns (where the discount includes a risk premium). So price to earnings is a measure of the amount of present investment dollars required to get an expected future return. So if PE goes up your expected future return on a dollar invested has gone down.

User avatar
HomerJ
Posts: 11564
Joined: Fri Jun 06, 2008 12:50 pm

Re: William Bernstein: the paradox of wealth

Post by HomerJ » Tue Oct 08, 2013 1:25 pm

thx1138 wrote:
Browser wrote: OK, now explain to me how PE10 is a measure of investment dollars per unit of return and I'll understand.
Well P is Price, and E is Earnings. Recent earnings are the best known ex ante indicator of future returns, though of course an imperfect indicator. And the dividend discount model estimates the current price as being the sum total of expected discounted future returns (where the discount includes a risk premium). So price to earnings is a measure of the amount of present investment dollars required to get an expected future return. So if PE goes up your expected future return on a dollar invested has gone down.
Good explanation...

Now explain why today's P/E ratio isn't used to determine future returns, but instead PE10 is used... Why not PE5 or PE15 or PE27.9... Why is PE10 so special? How do we know it can accurately predict the future? How far into the future can it predict? Next year? the next 5 years? The next 30 years?

exeunt
Posts: 907
Joined: Tue Jan 08, 2008 6:54 pm

Re: William Bernstein: the paradox of wealth

Post by exeunt » Tue Oct 08, 2013 1:38 pm

It doesn't matter which PE you use. They all pretty much say the same thing.

User avatar
stratton
Posts: 11082
Joined: Sun Mar 04, 2007 5:05 pm
Location: Puget Sound

Re: William Bernstein: the paradox of wealth

Post by stratton » Tue Oct 08, 2013 1:51 pm

Lower returns on investment can then lead to Minsky Moments.

Investment returns are so low leverage is increased to try and raise returns leading to BOOM!

Long Term Capital Management is an example as is 2007-2010, but more systemic.

Paul
...and then Buffy staked Edward. The end.

thx1138
Posts: 774
Joined: Fri Jul 12, 2013 2:14 pm

Re: William Bernstein: the paradox of wealth

Post by thx1138 » Tue Oct 08, 2013 1:56 pm

HomerJ wrote: Now explain why today's P/E ratio isn't used to determine future returns, but instead PE10 is used...
First understand that PE10 is current price divided by averaged earnings. It isn't PE averaged, only the earnings are averaged. This is to reduce the noise in earnings reporting. The current price is of course what matters on the price side since that is the price you will buy at, but earnings reporting is noisy so it is averaged. The PE typically reported is essentially PE1 (twelve month trailing earnings) and this is subject to short term fluctuations that do not correlate well with the overall health and earnings potential of the market. There is both theoretical and empirical evidence for this. One year earnings is too short a time window.
Why not PE5 or PE15 or PE27.9... Why is PE10 so special?
It doesn't matter much at all, again you just need some smoothing and PE5, PE10 and PE20 will all give similar indications. As to why ten, because in back testing it was reasonably predictive in the past - specifically of twenty year future returns. So a measure for those estimating long term returns - not short term day trader types.
How do we know it can accurately predict the future?
We don't of course, and anyone that thinks any ex ante measurement "predicts the future" is a fool. None of them do. The question is what is the best estimator to date. PE10 is better than many other options so it is commonly used. It is still a pretty poor estimator though. Which of course it has to be, otherwise there would be no risk premium.
How far into the future can it predict? Next year? the next 5 years? The next 30 years?
The data Schiller reported was for twenty year future returns.

Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Re: William Bernstein: the paradox of wealth

Post by Browser » Tue Oct 08, 2013 4:12 pm

thx1138 wrote:
Browser wrote: OK, now explain to me how PE10 is a measure of investment dollars per unit of return and I'll understand.
Well P is Price, and E is Earnings. Recent earnings are the best known ex ante indicator of future returns, though of course an imperfect indicator. And the dividend discount model estimates the current price as being the sum total of expected discounted future returns (where the discount includes a risk premium). So price to earnings is a measure of the amount of present investment dollars required to get an expected future return. So if PE goes up your expected future return on a dollar invested has gone down.
Well, I'll assume for the purposes of discussion that I understand this. But it doesn't seem to explain why the expected returns from stocks might be going down. I guess it has to do with the positive slope to the line over time, but there seem to be a lot of other possible explanations for that: one in particular is that the P/E bubble in the 1990s pulled up the slope, but that was probably a one-off event. Don't know for sure, but wouldn't the slope of the median values, instead of mean values, be a better indicator? The current median is 15.89. I don't know, but is that much different from what it has been historically?
We don't know where we are, or where we're going -- but we're making good time.

User avatar
HomerJ
Posts: 11564
Joined: Fri Jun 06, 2008 12:50 pm

Re: William Bernstein: the paradox of wealth

Post by HomerJ » Tue Oct 08, 2013 4:15 pm

thx1138 wrote:It doesn't matter much at all, again you just need some smoothing and PE5, PE10 and PE20 will all give similar indications.
Really? Will a PE5 that includes only the bull run from 2009-2014 give the same indication as a PE7 that includes the crash before the runup?

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

Re: William Bernstein: the paradox of wealth

Post by Rodc » Tue Oct 08, 2013 4:29 pm

Well, I'll assume for the purposes of discussion that I understand this. But it doesn't seem to explain why the expected returns from stocks might be going down.


Turn P/E10 upside down. A P/E10 of 15 means an E10/P of 6.7%. That is your earnings yield on your investment, or rather an estimate since the past average earnings may not be the future average earnings. In the long term, absent bubbles small or large or pops small or large (swings in P or E due to panics and giddyness) this is an estimate of the long term real return you might expect.

If over time the trend is towards higher P/E10, say from 15 to 20, that is a reduction in real return (expected) from 6.7% to 5%.

All standard caveats about estimates apply.
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
grayfox
Posts: 4932
Joined: Sat Sep 15, 2007 4:30 am

Re: William Bernstein: the paradox of wealth

Post by grayfox » Tue Oct 08, 2013 4:35 pm

You can't buy past earnings. When you buy stock, you are buying future earnings.

So I use a better measure for P/E10 than Shiller. Instead of averaging the past ten years real earnings, E10backward, I use the next 10 years of real earnings, E10forward. If you multiply E10forward by 10, then is the total earnings per share you will receive over the next ten years.

Using E10forward takes much of the uncertainty (and a lot of the fun) out of stock investing.

Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Re: William Bernstein: the paradox of wealth

Post by Browser » Tue Oct 08, 2013 4:43 pm

Thanks rodc. That's a lot clearer (I think). Is the PE10 trendline the regression line?
We don't know where we are, or where we're going -- but we're making good time.

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

Re: William Bernstein: the paradox of wealth

Post by Rodc » Tue Oct 08, 2013 6:00 pm

Browser wrote:Thanks rodc. That's a lot clearer (I think). Is the PE10 trendline the regression line?
Yes.

One problem is the slope of the trend line is hugely affected by the start and end dates. So it is hard to estimate the sense (plus or minus) of a trend line when the slope is small. That is if we get just one recession the whole method used might come to the exact opposite conclusion.
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
grayfox
Posts: 4932
Joined: Sat Sep 15, 2007 4:30 am

Re: William Bernstein: the paradox of wealth

Post by grayfox » Tue Oct 08, 2013 6:48 pm

This may be of interest: a chart showing E10backward (green line) that a Boglehead made in 2011.

Image
from http://www.bogleheads.org/forum/viewtop ... 32#p958332

You can also get E10forward just by shifting the line 10 years to the left.
For example, E10forward(Jan-1984) is the same as E10backward(Jan-1994)

Usually E10forward is larger than E10backward thanks to real earnings growth.
But it looks like not always.

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

Re: William Bernstein: the paradox of wealth

Post by Rodc » Tue Oct 08, 2013 8:08 pm

grok87 wrote:...
http://www.cfainstitute.org/learning/pr ... .n5.1.aspx
..
Another great article by Dr. Bernstein. He makes the case that stock and bond returns will be more muted going forward. He thinks fair value for Shillers PE10 may be 20 now instead of the historical average of 16.
Bernstein wrote: The familiar 10-year cyclically adjusted price-to-earning ratio (CAPE) series of US stocks...suggests that over the past 132 years earnings muliples have expanded with an average annual regression slope of 0.058.[But] the t-statstic for the 14 independent 10 year CAPE points is only 1.65, close but no cigar.
Here is the problem.

I just grabbed the data.

Indeed from the start in 1871 to today the slope of the regression line is 0.058.

However you can't just blindly compute statistics and understand things. You have to look at the data. It is pretty clear from the graph that this result is entirely driven by the great Tech Bubble.

In fact if you take out the last 20 years the trend line has a negative slope! If someone had done this study 20 years ago the story would have been the reverse! The 120-year story is one of P/E10 trending down, but the last 20 years and the great tech bubble throw everything so out of whack that suddenly the slope reverses.

So this is certainly not a story of some more or less nicely sloping up relationship over time. It is due to the by-far worst P/E10 bubble in history causing what is most likely an anomalous distortion in the data, not some fundamental truth about the markets.
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.

peppers
Posts: 1335
Joined: Tue Oct 25, 2011 7:05 pm

Re: William Bernstein: the paradox of wealth

Post by peppers » Tue Oct 08, 2013 8:21 pm

^Thank you, rodc.
"..the cavalry ain't comin' kid, you're on your own..."

Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Re: William Bernstein: the paradox of wealth

Post by Browser » Tue Oct 08, 2013 8:56 pm

I second. So then, what other data might there be that equity returns may be downtrending? If there isn't any, let's go with my opinion that we don't know.
We don't know where we are, or where we're going -- but we're making good time.

grok87
Posts: 8295
Joined: Tue Feb 27, 2007 9:00 pm

Re: William Bernstein: the paradox of wealth

Post by grok87 » Tue Oct 08, 2013 10:00 pm

Rodc wrote:
grok87 wrote:...
http://www.cfainstitute.org/learning/pr ... .n5.1.aspx
..
Another great article by Dr. Bernstein. He makes the case that stock and bond returns will be more muted going forward. He thinks fair value for Shillers PE10 may be 20 now instead of the historical average of 16.
Bernstein wrote: The familiar 10-year cyclically adjusted price-to-earning ratio (CAPE) series of US stocks...suggests that over the past 132 years earnings muliples have expanded with an average annual regression slope of 0.058.[But] the t-statstic for the 14 independent 10 year CAPE points is only 1.65, close but no cigar.
Here is the problem.

I just grabbed the data.

Indeed from the start in 1871 to today the slope of the regression line is 0.058.

However you can't just blindly compute statistics and understand things. You have to look at the data. It is pretty clear from the graph that this result is entirely driven by the great Tech Bubble.

In fact if you take out the last 20 years the trend line has a negative slope! If someone had done this study 20 years ago the story would have been the reverse! The 120-year story is one of P/E10 trending down, but the last 20 years and the great tech bubble throw everything so out of whack that suddenly the slope reverses.

So this is certainly not a story of some more or less nicely sloping up relationship over time. It is due to the by-far worst P/E10 bubble in history causing what is most likely an anomalous distortion in the data, not some fundamental truth about the markets.
thanks- good point.
it would be interesting to plot other countries to see if there is a similar pattern. not sure other countries are available back that far though...
Keep calm and Boglehead on. KCBO.

Bill Bernstein
Posts: 557
Joined: Sat Jun 23, 2007 12:47 am

Re: William Bernstein: the paradox of wealth

Post by Bill Bernstein » Wed Oct 09, 2013 12:09 am

Rodc:

It's possible to slice and dice the data in opposite direction, too, if you want: eliminate not only the 90s but the 70s as well, when valuations were anomalously (and illogically) low because of high inflation, and the slope becomes positive again. Eliminate the Depression, and it becomes even more positive.

That said, as I mention in the paper, the t-stat for 14 independent 10-year periods is only 1.6, so the case isn't proven for large cap US stocks.

Bill

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

Re: William Bernstein: the paradox of wealth

Post by Rodc » Wed Oct 09, 2013 6:22 am

wbern wrote:Rodc:

It's possible to slice and dice the data in opposite direction, too, if you want: eliminate not only the 90s but the 70s as well, when valuations were anomalously (and illogically) low because of high inflation, and the slope becomes positive again. Eliminate the Depression, and it becomes even more positive.

That said, as I mention in the paper, the t-stat for 14 independent 10-year periods is only 1.6, so the case isn't proven for large cap US stocks.

Bill
Hi Bill,

Sure, lots of playing could be done, but one would want a clear rationale for doing so. And of course the more one does that the more suspect the results. The Great Tech bubble was of course real and should not be entirely discounted. However one anomalous data point or time period does not make a "predictive" trend. That is a trend with any deeper meaning and more importantly a trend that one should expect to continue.

I would also note that the slope of a trend line is greatly driven by the data near the ends of the time period and the data in the middle matter much less (just an effect of the length of a lever and that nature of squaring the distance from than lever to the data), which not infrequently causes problems like this. So it is always good to think about whether the beginning and ending data are "representative". Data anomalies or outliers are most damaging to the results if they lie near the beginning or as in this case, near the end. In this case it makes the Great Tech bubble a double whammy as far as throwing off the results.

it is always extremely important to understand what drives a particular statistical result because just as in this example the reason may help one to understand if the result is important and useful or not. Just because one can show a correlation between butter prices in Bangladesh and the stock market (a famous example) does not make the result meaningful. You did mention the result is not statistically significant, which is not surprising as modest changes to the data change the results rather up or down, which is not surprising from a look at a graph of the data.

So from both a lack of rigorous statistical basis and from the obvious bias introduced by the Tech Bubble I would suggest this is really a non-issue as far as one's approach to investing.

Best,
Rod
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
grayfox
Posts: 4932
Joined: Sat Sep 15, 2007 4:30 am

Re: William Bernstein: the paradox of wealth

Post by grayfox » Wed Oct 09, 2013 8:51 am

If you are looking for a trend, you can find one in E10. (green line)

Image
from http://www.bogleheads.org/forum/viewtop ... 32#p958332

Earnings are Sales * ProfitMargin. ProfitMargin fluctuates but can not have a long-term trend. Sales, on the other hand, have a long term-up trend. Sales for the aggregate market is GDP, which has an up trend due to population growth and productivity improvements.

Now Poster wbern has a theory that stock returns have a long-term downward trend, which is the same as saying valuations have a long-term up trend. Poster rodc looked at the data and doesn't confirm an uptrend.

Back in 2010 I looked at how AVERAGE(P/E10) varied over time. The motivation was the notion of mean reversion of P/E10, so I was curious about which mean value P/E10 was supposed to revert to.

Image

Closer look at AVERAGE(P/E10):
Image

In 1995, we would have said that the mean of P/E10 was about 14.5. IN 2010, we would have said it was above 16.

One thing I observe is that, as more more samples are added, the sample mean is not converging to any particular mean. So the process is probably not stationary. In other words, statistics like mean and variance vary over time.

But I also don't observe any clear trend either up or down in P/E10. The mean has drifted randomly between about 14 and 18. Since about 1995, the sample mean has been rising, and will continue to rise as long as P/E10 is above average. Does that make a trend?
Last edited by grayfox on Wed Oct 09, 2013 8:55 am, edited 1 time in total.

Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Re: William Bernstein: the paradox of wealth

Post by Browser » Wed Oct 09, 2013 8:54 am

So, what data points were used to compute "average PE10"? PE10 is already a 10-year average of PEs, so do the data points represent an "average of average PEs". If so, then it seems like what you did was to smooth the average even more by turning it into a PE20 or PE30 or something like that.
We don't know where we are, or where we're going -- but we're making good time.

User avatar
grayfox
Posts: 4932
Joined: Sat Sep 15, 2007 4:30 am

Re: William Bernstein: the paradox of wealth

Post by grayfox » Wed Oct 09, 2013 8:58 am

Browser wrote:So, what data points were used to compute "average PE10"? PE10 is already a 10-year average of PEs, so do the data points represent an "average of average PEs".
P/E10 is NOT a 10-year average of PEs. You need to get that straight first.

AVERAGE(P/E10) is the sample mean, The blue line shows what an observer would have said the mean of P/E10 was though time.

Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Re: William Bernstein: the paradox of wealth

Post by Browser » Wed Oct 09, 2013 9:00 am

grayfox wrote:
Browser wrote:So, what data points were used to compute "average PE10"? PE10 is already a 10-year average of PEs, so do the data points represent an "average of average PEs".
PE10 is NOT a 10-year average of PEs. You need to get that straight first.

AVERAGE(P/E10) is the sample mean, The blue line shows what an observer would have said the mean of P/E10 was though time.
Not sure I understand you. I thought that PE10 was a 10-year moving average of real PE. What is the "sample" you are referring to?
We don't know where we are, or where we're going -- but we're making good time.

User avatar
grayfox
Posts: 4932
Joined: Sat Sep 15, 2007 4:30 am

Re: William Bernstein: the paradox of wealth

Post by grayfox » Wed Oct 09, 2013 9:02 am

Browser wrote:
grayfox wrote:
Browser wrote:So, what data points were used to compute "average PE10"? PE10 is already a 10-year average of PEs, so do the data points represent an "average of average PEs".
PE10 is NOT a 10-year average of PEs. You need to get that straight first.

AVERAGE(P/E10) is the sample mean, The blue line shows what an observer would have said the mean of P/E10 was though time.
Not sure I understand you. I thought that PE10 was a 10-year moving average of real PE.
That is a common misunderstanding. E10 is the 10-year moving average of E. Then you take the current price P and divide it by E10. P/E10.

That's not the same is taking each of the last 10 years P/E1, adding them up, and dividing by 10.

Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Re: William Bernstein: the paradox of wealth

Post by Browser » Wed Oct 09, 2013 9:08 am

grayfox wrote:
Browser wrote:
grayfox wrote:
Browser wrote:So, what data points were used to compute "average PE10"? PE10 is already a 10-year average of PEs, so do the data points represent an "average of average PEs".
PE10 is NOT a 10-year average of PEs. You need to get that straight first.

AVERAGE(P/E10) is the sample mean, The blue line shows what an observer would have said the mean of P/E10 was though time.
Not sure I understand you. I thought that PE10 was a 10-year moving average of real PE.
That is a common misunderstanding. E10 is the 10-year moving average of E. Then you take the current price P and divide it by E10. P/E10.

That's not the same is taking each of the last 10 years P/E1, adding them up, and dividing by 10.
Aha! Can you explain exactly how your sample mean PE10 is computed?
We don't know where we are, or where we're going -- but we're making good time.

User avatar
grayfox
Posts: 4932
Joined: Sat Sep 15, 2007 4:30 am

Re: William Bernstein: the paradox of wealth

Post by grayfox » Wed Oct 09, 2013 9:21 am

Browser wrote:
Aha! Can you explain exactly how your sample mean PE10 is computed?[/quote]

Go to http://www.multpl.com/shiller-pe/ which shows are plot of P/E10 from 1881 to 2013. The current value of P/E10 = 23.69. Down in the lower left it shows

Mean = 16.49
Median = 15.89
Min = 4.78 (Dec 1920)
Max = 44.20 (Dec 1999)

How did they calculate the mean? Just take all the P/E10's for each year (or month), add them up, and dividend by the number of years. It's just the arithmetic average. In 2013, the mean works out to 16.49.

Now if you calculated the mean back in 1995, it would have been about 14.5.

If the process were stationary, as n increases, the sample mean should converge to the population mean. So over time the estimate of the mean would get better and better.

To me, it does not look the mean is converging to any particular value. wbern has a theory that P/E10 is trending up. Maybe in 100 years, mean P/E10 will be over 20.

Bill Bernstein
Posts: 557
Joined: Sat Jun 23, 2007 12:47 am

Re: William Bernstein: the paradox of wealth

Post by Bill Bernstein » Wed Oct 09, 2013 10:33 am

You have to take the PE10 plot in the context of the larger article, which points out that there definitely has been a decrease in the cost of capital over the millennia, and it's a very noisy process.

So noisy, in fact, that you can't really be sure you're seeing it in the Shiller data.

As this thread well demonstrates.

Bill

User avatar
Slick8503
Posts: 195
Joined: Tue Nov 16, 2010 10:25 am

Re: William Bernstein: the paradox of wealth

Post by Slick8503 » Wed Oct 09, 2013 10:34 am

What about throwing out the highest peak(90's tech bubble) and the lowest valley? (great depression?)

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

Re: William Bernstein: the paradox of wealth

Post by Rodc » Wed Oct 09, 2013 1:30 pm

wbern wrote:You have to take the PE10 plot in the context of the larger article, which points out that there definitely has been a decrease in the cost of capital over the millennia, and it's a very noisy process.

So noisy, in fact, that you can't really be sure you're seeing it in the Shiller data.

As this thread well demonstrates.

Bill
Fair enough.
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.

Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Re: William Bernstein: the paradox of wealth

Post by Browser » Wed Oct 09, 2013 9:18 pm

How did they calculate the mean? Just take all the P/E10's for each year (or month), add them up, and dividend by the number of years. It's just the arithmetic average. In 2013, the mean works out to 16.49.

Now if you calculated the mean back in 1995, it would have been about 14.5.

If the process were stationary, as n increases, the sample mean should converge to the population mean. So over time the estimate of the mean would get better and better.

To me, it does not look the mean is converging to any particular value. wbern has a theory that P/E10 is trending up. Maybe in 100 years, mean P/E10 will be over 20.
What seems a little strange to me is that each mean value data point subsumes the prior mean values. It effectively places the highest weight on the oldest mean value and the least weight on the most recent mean value. For example, let's assume there is a series of four data points: 2, 4, 6, 8. The first mean is (2+4)/2 = 3. The second mean is the ((first mean (3) * 2) + 6)/3 = (6+6)/3 = 4. The third mean is the ((second mean (4) * 3) + 8)/ 4 = 20/4 = 5, which subsumes the first mean, and so forth. So it's not terribly surprising that the seqential sample size mean values would wander around in a narrow range that doesn't show any trend, is it?
We don't know where we are, or where we're going -- but we're making good time.

MIretired
Posts: 733
Joined: Fri Sep 06, 2013 12:35 pm

Re: William Bernstein: the paradox of wealth

Post by MIretired » Wed Oct 09, 2013 10:01 pm

2, 4, 6, 8. The first mean is (2+4)/2 = 3. The second mean is the ((first mean (3) * 2) + 6)/3 = (6+6)/3 = 4. The third mean is the ((second mean (4) * 3) + 8)/ 4 = 20/4 = 5
Each year gets a weight of 1. [(past 3 years avg * 3) + new yr.]/4 = 4 points averaged,[( the last 3 avg * 3) + new]/4

Post Reply