Valuation-based market timing with PE10 can improve returns?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
jln
Posts: 178
Joined: Thu Dec 09, 2010 2:18 pm

Goyal and Welch Paper

Post by jln » Tue Jun 21, 2011 4:29 pm

This topic has been studied before, notably in a 2004 academic paper by Amit Goyal of Emory University and Ivo Welch of Yale University. Here's a link to a PDF version of their paper:

A Comprehensive Look at the Empirical Performance of Equity Premium Prediction

Here's the abstract:
Given the historically high equity premium, is it now a good time to invest in the stock market? Economists have suggested a whole range of variables that investors could or should use to predict: dividend price ratios, dividend yields, earnings-price ratios, dividend payout ratios, net issuing ratios, book-market ratios, interest rates (in various guises), and consumption- based macroeconomic ratios (cay). The typical paper reports that the variable predicted well in an in-sample regression, implying forecasting ability.
Our paper explores the out-of-sample performance of these variables, and finds that not a single one would have helped a real-world investor outpredicting the then-prevailing historical equity premium mean. Most would have outright hurt. Therefore, we find that, for all practical purposes, the equity premium has not been predictable, and any belief about whether the stock market is now too high or too low has to be based on theoretical prior, not on the empirically variables we have explored.
Among many other variables and variations of various strategies, they investigated PE10, which is the strategy being discussed in this thread. They report their results in section 4.3 on page 18:
Lamont (1998) explores variations of the E/P ratio and the payout ratio. Table 8 thus explores variations on the computation of earnings and dividend ratios. For example, Earning(10Y) are the moving average 10-year earnings. We explore two different horizons: one in which the forecast begins in 1902, another in which the forecast begins in 1964.

Panel A shows that there is both statistically significant outperformance (both in-sample and out-of-sample!) the price variables (e/p and d/p) if we use longer-term moving average price ratios and if we begin our forecasts in 1902. The economic significance reached 33 basis points for the Earnings(10Y)/Price ratio, though it is below 10 basis points for all other variations. The payout ratio (d/e) does not work. Unfortunately, if we begin our forecasts in 1964, it is not just that our variables are no longer statistically significant, they outright underperform.

Panels B through D explore 3-year, 5-year, and 10-year horizons, respectively. Panel B shows that the 3-year horizon predictions look like the one-year horizons: significant outperformance of long-memory price ratios if we begin in 1902, but underperformance if we begin in 1964. Panel C and D show that the 5-year and 10-year horizon predictions become progressively worse. The 10-year horizon predictions, however, show statistically insignificant overperformance when we begin predictions in 1964, but none if we begin predictions in 1902.

In sum, there is a tiny hint that long-memory earnings-price ratios might have better in-sample and out-of-sample performance than the prevailing mean; but the empirical evidence is so modest that it is better interpreted as not speaking against e/p, instead of speaking in favor for e/p. It looks decent primarily because the other predictive variables look so incredibly bad.
Note that the "earnings-price" ratio (EP) used by Goyal and Welch is simply the inverse of the "price-to-earnings" ratio (PE). So the results they found for EP10 apply equally well to PE10. See the paper for the tables and charts referenced in the quote above, as well as for details on methodology, other results, etc.

So Goyal and Welch are not as enthusiastic about using PE10 as are the authors of the paper discussed here. Make of this what you will. I believe that it remains a rather contentious topic in the academic world, with respected researchers on both sides of the debate. I'm not competent to judge all of this, but it is rather interesting, and if anyone is still following this long thread, I thought you might like to see a contrary point of view.

John Norstad

User avatar
fredflinstone
Posts: 2101
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone » Tue Jun 21, 2011 4:37 pm

rrosenkoetter wrote:
tadamsmar wrote:On page 5 he says:
More research is needed about this.
If you are going to commit to this stuff, you need to hope that he was not right about that!
fred's already rewritten his IPS... uh-oh.
no i haven't.

User avatar
fredflinstone
Posts: 2101
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone » Tue Jun 21, 2011 4:40 pm

rrosenkoetter wrote:
fredflinstone wrote:I don't accept the premise that the "Stay the Course" philosophy precludes valuation-based market-timing. To my mind, "Stay the Course" simply means writing an IPS and sticking to it. If my IPS tells me to reduce equity allocation if PE10 reaches say 40, then Staying the Course means I should reduce equity allocation at a PE of 40. At least that's how I see it.
Until you read some other study 5 years from now that offers "even better returns with less risk!!! With proof!", and you'll change your IPS again.

Good luck.
My position on all this is similar to that of Jack Bogle, who advocates "gentle" market timing at extreme valuations. When you ridicule me, you are also ridiculing Mr. Bogle.

letsgobobby
Posts: 11421
Joined: Fri Sep 18, 2009 1:10 am

Post by letsgobobby » Tue Jun 21, 2011 5:35 pm

also Ben Graham, Warren Buffett, Charles Ellis, Larry Swedroe, Bill Bernstein, and a few others I'm sure I've forgotten, all of whom in one way or another said in the late 90s that stocks were overvalued and likely to have modest long term returns from there; or at least that there are times when stock markets are overvalued and holdings may be reduced. This is not a new idea; nor does it seem particularly controversial but I am clearly wrong on the latter point.
Last edited by letsgobobby on Tue Jun 21, 2011 6:34 pm, edited 1 time in total.

User avatar
tadamsmar
Posts: 7725
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar » Tue Jun 21, 2011 5:53 pm

novastepp wrote:Does the study take into account a strict rebalancing strategy that many Bogelheads use; say a 5% deviation or an annual effort?

I see that the study crosses its numbers with buy and hold, but I do not see mention as to the always-forgotten rebalancing strategy.
That study does not even take into account the high capital gains rates that investors would have to pay in the historical back test period. Correct that, it does not even deserve to be called a back test. The author is not advocating that anyone use this as an investment strategy, he is just advocating more research to turn it into a realistic back test.

Most of the posters on this thread need to give up the wishful thinking and actually start reading these papers.

User avatar
tadamsmar
Posts: 7725
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar » Tue Jun 21, 2011 6:09 pm

rrosenkoetter wrote:
peter71 wrote:
rrosenkoetter wrote:
wade wrote:Then you just have to commit yourself to sticking to it no matter what, which means you should take some time and explore the possibilities and make sure you won't be scared away from following through with your plan.
LOL... Yep, based on this study, you'll have to commit to sticking with this new plan for the next 30 years no matter how bad it may seem to be performing at times...

and hope that Dr. Pfau was right.

Dr Pfau... you realize that your studies may impact people lives, right?

Be very careful in your ivory tower.
Give it a rest. Dr. Pfau and I suspect most participants in this thread understand the case you're trying to make much better than you understand it yourself.
Oh, are we the only people on the Internet who have access to and have ever read this study? He isn't just presenting the data... He's pushing using it as a trading strategy.
At one point he calls for more research, since he left out the historical tax rates that, of course, must be included in a realistic back test.

But in the conclusion he forgets about this and seems to start advocating actually using this without more research! Opps!

Pfau can't write a paper without contradicting himself. It appears to be an unpublished draft. If he ever gets it published, the reviewers will make him tone down the conclusions to be consistent with the rest if the paper.

Beware, gullible ones, poor wishful thinkers.

chaster86
Posts: 194
Joined: Mon May 16, 2011 12:15 pm

Post by chaster86 » Tue Jun 21, 2011 6:09 pm

Ok you convinced me. I've decided I will allocate a portion of my portfolio to the strategies advocated by fredflinstone, letsgobobby, Bogle, Graham, Buffett, Ellis, Swedroe, and Berstein. After all, the numbers clearly show that if only 10% of investors did this from 1871 then those investors would be able to purchase every outstanding share on the stock market today! Impressive!

Now, what portion shall I allocate? I will allocate to each in proportion to the amount of assets they've acquired. Surely if it's the path to riches then these investors, along with their disciples will amass rather large portfolios and come to dominate the market (10%!). So if they're right I will share in their profits and if they're wrong then I'll eschew their loses.

How does that sound?

letsgobobby
Posts: 11421
Joined: Fri Sep 18, 2009 1:10 am

Post by letsgobobby » Tue Jun 21, 2011 7:59 pm

Someone please tell me what the risk is of following PE10. My understanding is there are 2 risks:

1. Underperforming on the upside due to underexposure to stocks because PE10 is high. I can live with this risk. At worst, I have reduced returns with reduced volatility. I have a low need for risk.

2. In the event PE10 is low and goes substantially lower and stays low indefinitely, I would underperform to the downside, and with greater volatility. Although I think this is unlikely over a 20 year period of time, I acknowledge it as a risk and can live with it.

Was there anything else?

User avatar
fredflinstone
Posts: 2101
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone » Tue Jun 21, 2011 8:28 pm

tadamsmar wrote:
novastepp wrote:Does the study take into account a strict rebalancing strategy that many Bogelheads use; say a 5% deviation or an annual effort?

I see that the study crosses its numbers with buy and hold, but I do not see mention as to the always-forgotten rebalancing strategy.
That study does not even take into account the high capital gains rates that investors would have to pay in the historical back test period. Correct that, it does not even deserve to be called a back test. The author is not advocating that anyone use this as an investment strategy, he is just advocating more research to turn it into a realistic back test.

Most of the posters on this thread need to give up the wishful thinking and actually start reading these papers.
I can't speak for Professor Pfau, but I care not one whit about historical capital gains rates. The tax rate in 1898 or 1921 or whatever has no bearing on my investment strategy. All that matters to me are current and future tax rates.

User avatar
fredflinstone
Posts: 2101
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone » Tue Jun 21, 2011 8:29 pm

letsgobobby wrote:also Ben Graham, Warren Buffett, Charles Ellis, Larry Swedroe, Bill Bernstein, and a few others I'm sure I've forgotten, all of whom in one way or another said in the late 90s that stocks were overvalued and likely to have modest long term returns from there; or at least that there are times when stock markets are overvalued and holdings may be reduced. This is not a new idea; nor does it seem particularly controversial but I am clearly wrong on the latter point.
Not only that, but some of the above (plus Bogle) advocate various forms of market-timing with respect to bonds or TIPS. They may not call it market-timing, but in my view, that is what it is.

User avatar
fredflinstone
Posts: 2101
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone » Tue Jun 21, 2011 8:32 pm

letsgobobby wrote:Someone please tell me what the risk is of following PE10. My understanding is there are 2 risks:

1. Underperforming on the upside due to underexposure to stocks because PE10 is high. I can live with this risk. At worst, I have reduced returns with reduced volatility. I have a low need for risk.

2. In the event PE10 is low and goes substantially lower and stays low indefinitely, I would underperform to the downside, and with greater volatility. Although I think this is unlikely over a 20 year period of time, I acknowledge it as a risk and can live with it.

Was there anything else?
I cannot think of anything. By the way, this resonates with me: "At worst, I have reduced returns with reduced volatility. I have a low need for risk."

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

Post by Rodc » Tue Jun 21, 2011 9:12 pm

fredflinstone wrote:
tadamsmar wrote:
novastepp wrote:Does the study take into account a strict rebalancing strategy that many Bogelheads use; say a 5% deviation or an annual effort?

I see that the study crosses its numbers with buy and hold, but I do not see mention as to the always-forgotten rebalancing strategy.
That study does not even take into account the high capital gains rates that investors would have to pay in the historical back test period. Correct that, it does not even deserve to be called a back test. The author is not advocating that anyone use this as an investment strategy, he is just advocating more research to turn it into a realistic back test.

Most of the posters on this thread need to give up the wishful thinking and actually start reading these papers.
I can't speak for Professor Pfau, but I care not one whit about historical capital gains rates. The tax rate in 1898 or 1921 or whatever has no bearing on my investment strategy. All that matters to me are current and future tax rates.
It should matter to you.

People at that time priced stocks (as best they could) to give a decent return for the risk. Returns net of taxes, trading costs, etc.

When tax and other drag are high, stocks are priced to give a higher expected return to get an ok net return (for the risk).

So if you only look at returns without considering past drag you have very unrealistic values. In particular returns you can't expect to get now unless the drag is the same.

Since the drag changes over time, this is just another reason why Past Returns do not Predict Future Returns very well.

This is another one of those cases where you can't just use numbers blindly without really understanding their limitations.
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
fredflinstone
Posts: 2101
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone » Wed Jun 22, 2011 5:18 am

Rodc wrote:
fredflinstone wrote:
tadamsmar wrote:
novastepp wrote:Does the study take into account a strict rebalancing strategy that many Bogelheads use; say a 5% deviation or an annual effort?

I see that the study crosses its numbers with buy and hold, but I do not see mention as to the always-forgotten rebalancing strategy.
That study does not even take into account the high capital gains rates that investors would have to pay in the historical back test period. Correct that, it does not even deserve to be called a back test. The author is not advocating that anyone use this as an investment strategy, he is just advocating more research to turn it into a realistic back test.

Most of the posters on this thread need to give up the wishful thinking and actually start reading these papers.
I can't speak for Professor Pfau, but I care not one whit about historical capital gains rates. The tax rate in 1898 or 1921 or whatever has no bearing on my investment strategy. All that matters to me are current and future tax rates.
It should matter to you.

People at that time priced stocks (as best they could) to give a decent return for the risk. Returns net of taxes, trading costs, etc.

When tax and other drag are high, stocks are priced to give a higher expected return to get an ok net return (for the risk).

So if you only look at returns without considering past drag you have very unrealistic values. In particular returns you can't expect to get now unless the drag is the same.

Since the drag changes over time, this is just another reason why Past Returns do not Predict Future Returns very well.

This is another one of those cases where you can't just use numbers blindly without really understanding their limitations.
That is an excellent point. I hadn't considered it. Thank you. I retract my statement that I don't care about historical tax rates. It would be good to see Pfau's analysis both ways--once including analysis of taxes and once without.

Because (a) the over-performance of the market-timing approach is large and (b) the market-timing approach does not require much trading, I am confident that market-timing will prevail over standard rebalancing in both cases. But it would be interesting to see how much the margin decreases when taxes are taken into account.

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

Post by Rodc » Wed Jun 22, 2011 6:37 am

Because (a) the over-performance of the market-timing approach is large and (b) the market-timing approach does not require much trading, I am confident that market-timing will prevail over standard rebalancing in both cases. But it would be interesting to see how much the margin decreases when taxes are taken into account.
Points a and b are certainly not a given. I have played with the data a fair amount and I would argue it is not generally large or consistent.

Here is one such look I posted above:

http://home.comcast.net/~rodec/finance/ ... mingV2.pdf

On the common sense front: if it truly was a big consistent win (your points a and b taken together) it would be more widely used. This is not some new idea. In fact I have some hope that enough people now use P/E10 and other valuation metrics that it helps damp down highs and lows and could be part of why we never hit single digit lows in the last crash (but we likely will never know).

To the extent that you do not trade often, use a sliding scale without huge shifts, do it in a 401K or similar account with very low costs and no taxes I don't think this has huge risks. But if you do those things I also do not see much likelihood of large wins. Really the only way to have returns very different from a buy, hold, rebalance portfolio is to have a strategy that is very different: a portfolio with a slow modest shift up or down on stocks from BHR is just not all that much different from BHR to matter much. To soundly beat BHR you have to make a bold departure. If you do that you join the ranks of folks who will succeed big or lose big (all of whom think they have hit on the winning strategy). Indeed, the other big world beating strategy getting lots of press on this board is moving average market timing. How do you decide which group of smart world beaters is correct?

Just yesterday you learned something new about returns data. What will you learn tomorrow or next year that is new? How can you be so confident you are in the group with the great idea vs the group that says "whoops!"?

As long as you stick to something that is not wildly different from BHR you will likely do OK, so I'm not going to give you the "You are going to crash, burn and die!" speech. :) But if you expect great things from modest changes to BHR I think you will be disappointed. If you choose to make wild changes, well, you might just get a little wilder ride. Only time will tell.

Unfortunately, for all the dogma in investing on all sides, there is darn little we know with any certainty. s

Sincerely, best of luck, I wish you well in your investing.
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
tadamsmar
Posts: 7725
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar » Wed Jun 22, 2011 7:19 am

fredflinstone wrote:
Rodc wrote:
fredflinstone wrote:
tadamsmar wrote:
novastepp wrote:Does the study take into account a strict rebalancing strategy that many Bogelheads use; say a 5% deviation or an annual effort?

I see that the study crosses its numbers with buy and hold, but I do not see mention as to the always-forgotten rebalancing strategy.
That study does not even take into account the high capital gains rates that investors would have to pay in the historical back test period. Correct that, it does not even deserve to be called a back test. The author is not advocating that anyone use this as an investment strategy, he is just advocating more research to turn it into a realistic back test.

Most of the posters on this thread need to give up the wishful thinking and actually start reading these papers.
I can't speak for Professor Pfau, but I care not one whit about historical capital gains rates. The tax rate in 1898 or 1921 or whatever has no bearing on my investment strategy. All that matters to me are current and future tax rates.
It should matter to you.

People at that time priced stocks (as best they could) to give a decent return for the risk. Returns net of taxes, trading costs, etc.

When tax and other drag are high, stocks are priced to give a higher expected return to get an ok net return (for the risk).

So if you only look at returns without considering past drag you have very unrealistic values. In particular returns you can't expect to get now unless the drag is the same.

Since the drag changes over time, this is just another reason why Past Returns do not Predict Future Returns very well.

This is another one of those cases where you can't just use numbers blindly without really understanding their limitations.
That is an excellent point. I hadn't considered it. Thank you. I retract my statement that I don't care about historical tax rates. It would be good to see Pfau's analysis both ways--once including analysis of taxes and once without.

Because (a) the over-performance of the market-timing approach is large and (b) the market-timing approach does not require much trading, I am confident that market-timing will prevail over standard rebalancing in both cases. But it would be interesting to see how much the margin decreases when taxes are taken into account.
Unlike you, I agree with Pfau that more research is needed.

User avatar
fredflinstone
Posts: 2101
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone » Wed Jun 22, 2011 7:42 am

tadamsmar wrote:
fredflinstone wrote:
Rodc wrote:
fredflinstone wrote:
tadamsmar wrote: That study does not even take into account the high capital gains rates that investors would have to pay in the historical back test period. Correct that, it does not even deserve to be called a back test. The author is not advocating that anyone use this as an investment strategy, he is just advocating more research to turn it into a realistic back test.

Most of the posters on this thread need to give up the wishful thinking and actually start reading these papers.
I can't speak for Professor Pfau, but I care not one whit about historical capital gains rates. The tax rate in 1898 or 1921 or whatever has no bearing on my investment strategy. All that matters to me are current and future tax rates.
It should matter to you.

People at that time priced stocks (as best they could) to give a decent return for the risk. Returns net of taxes, trading costs, etc.

When tax and other drag are high, stocks are priced to give a higher expected return to get an ok net return (for the risk).

So if you only look at returns without considering past drag you have very unrealistic values. In particular returns you can't expect to get now unless the drag is the same.

Since the drag changes over time, this is just another reason why Past Returns do not Predict Future Returns very well.

This is another one of those cases where you can't just use numbers blindly without really understanding their limitations.
That is an excellent point. I hadn't considered it. Thank you. I retract my statement that I don't care about historical tax rates. It would be good to see Pfau's analysis both ways--once including analysis of taxes and once without.

Because (a) the over-performance of the market-timing approach is large and (b) the market-timing approach does not require much trading, I am confident that market-timing will prevail over standard rebalancing in both cases. But it would be interesting to see how much the margin decreases when taxes are taken into account.
Unlike you, I agree with Pfau that more research is needed.
?? I just said more research would be "interesting" and "good." To assist the reader, I have highlighted these statements in bold and red above. It is annoying when people deliberately mischaracterize what I write. If you want to criticize my arguments, fine, but please don't criticize things I didn't say.

User avatar
fredflinstone
Posts: 2101
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone » Wed Jun 22, 2011 7:50 am

Rodc wrote:
Because (a) the over-performance of the market-timing approach is large and (b) the market-timing approach does not require much trading, I am confident that market-timing will prevail over standard rebalancing in both cases. But it would be interesting to see how much the margin decreases when taxes are taken into account.
Points a and b are certainly not a given. I have played with the data a fair amount and I would argue it is not generally large or consistent.

Here is one such look I posted above:

http://home.comcast.net/~rodec/finance/ ... mingV2.pdf

On the common sense front: if it truly was a big consistent win (your points a and b taken together) it would be more widely used. This is not some new idea. In fact I have some hope that enough people now use P/E10 and other valuation metrics that it helps damp down highs and lows and could be part of why we never hit single digit lows in the last crash (but we likely will never know).

To the extent that you do not trade often, use a sliding scale without huge shifts, do it in a 401K or similar account with very low costs and no taxes I don't think this has huge risks. But if you do those things I also do not see much likelihood of large wins. Really the only way to have returns very different from a buy, hold, rebalance portfolio is to have a strategy that is very different: a portfolio with a slow modest shift up or down on stocks from BHR is just not all that much different from BHR to matter much. To soundly beat BHR you have to make a bold departure. If you do that you join the ranks of folks who will succeed big or lose big (all of whom think they have hit on the winning strategy). Indeed, the other big world beating strategy getting lots of press on this board is moving average market timing. How do you decide which group of smart world beaters is correct?

Just yesterday you learned something new about returns data. What will you learn tomorrow or next year that is new? How can you be so confident you are in the group with the great idea vs the group that says "whoops!"?

As long as you stick to something that is not wildly different from BHR you will likely do OK, so I'm not going to give you the "You are going to crash, burn and die!" speech. :) But if you expect great things from modest changes to BHR I think you will be disappointed. If you choose to make wild changes, well, you might just get a little wilder ride. Only time will tell.

Unfortunately, for all the dogma in investing on all sides, there is darn little we know with any certainty. s

Sincerely, best of luck, I wish you well in your investing.
Thanks. I am still learning. I expect I will continue to learn about investing and I am open to changing my IPS as I do so. To ignore new information seems to me to be foolish. The strategy outlined by Pfau provided much higher risk-adjusted returns over the last century than did the standard buy-hold-rebalance approach. Ignoring valuations seems to me to be foolish.

The position I am advocating is not identical to the one Pfau modeled in his paper. Rather, my position is basically the same as that advocated by Jack Bogle: buy-hold-and-rebalance for most of the portfolio but allow some "gentle" market timing at the margins when valuations are extreme. As letsgobobby has pointed out, lots of respected market gurus advocate this approach. So perhaps it is more common than you think, albeit difficult to carry out in practice due to emotions.

User avatar
tadamsmar
Posts: 7725
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar » Wed Jun 22, 2011 7:55 am

fredflinstone wrote:
tadamsmar wrote:
fredflinstone wrote:
Rodc wrote:
fredflinstone wrote: I can't speak for Professor Pfau, but I care not one whit about historical capital gains rates. The tax rate in 1898 or 1921 or whatever has no bearing on my investment strategy. All that matters to me are current and future tax rates.
It should matter to you.

People at that time priced stocks (as best they could) to give a decent return for the risk. Returns net of taxes, trading costs, etc.

When tax and other drag are high, stocks are priced to give a higher expected return to get an ok net return (for the risk).

So if you only look at returns without considering past drag you have very unrealistic values. In particular returns you can't expect to get now unless the drag is the same.

Since the drag changes over time, this is just another reason why Past Returns do not Predict Future Returns very well.

This is another one of those cases where you can't just use numbers blindly without really understanding their limitations.
That is an excellent point. I hadn't considered it. Thank you. I retract my statement that I don't care about historical tax rates. It would be good to see Pfau's analysis both ways--once including analysis of taxes and once without.

Because (a) the over-performance of the market-timing approach is large and (b) the market-timing approach does not require much trading, I am confident that market-timing will prevail over standard rebalancing in both cases. But it would be interesting to see how much the margin decreases when taxes are taken into account.
Unlike you, I agree with Pfau that more research is needed.
?? I just said more research would be "interesting" and "good." To assist the reader, I have highlighted these statements in bold and red above. It is annoying when people deliberately mischaracterize what I write. If you want to criticize my arguments, fine, but please don't criticize things I didn't say.
You said: I am confident that market-timing will prevail over standard rebalancing in both cases.

By "prevail", I guess you just mean show returns in a backtest. Or do you mean make you richer than than the Boglehead approach?

User avatar
DRiP Guy
Posts: 2237
Joined: Tue Feb 20, 2007 4:54 pm

Post by DRiP Guy » Wed Jun 22, 2011 8:02 am

fredflinstone wrote:To ignore new information seems to me to be foolish. The strategy outlined by Pfau provided much higher risk-adjusted returns over the last century than did the standard buy-hold-rebalance approach. Ignoring valuations seems to me to be foolish.
With a maximum of respect for both you and the work you cite, failing to instantly adopt some new draft proposal, based on a freshman effort in a highly controversial and difficult to predict area of financial planning, is hardly the same as just blithely ignoring information.

For you to both assume, and then to twice rapidly and vocally assert as much, is really rather rude, IMHO.

I will, for your benefit, once again share my own view on 'valuations'*, a view I will not be so bold as to claim represents the overall Boglehead view, but one I do feel more than one Boglehead likely carries:

"Valuations are important to determining future returns. Thus it has always been, since long before Graham and Dodd went on at length about it some 70 years ago, now. It has always been what to do about them, if anything, that is problematic for the individual investor.

For that reason, taking a long-term approach, buying a broad market basket, diversifying, managing risk, and establishing (then maintaining) an AA that is in line with all of that, is an approach that removes the danger of trying to guess the 'proper' valuation of the market, and also what that may bode for it's future, then to capitalize on that by being able to TIME trading into and/or out of the market, based on the valuation assessment. "

Whew.



*As defined by whatever metric, metrics, or other current definition might apply. Personally, I think that is one of the many, but mostly unstated, follies of any valuation-based approach to timing your trading -- deciding on what a 'proper' valuation really is, given the prevailing (and future) conditions that will influence it.

User avatar
Mel Lindauer
Moderator
Posts: 28400
Joined: Mon Feb 19, 2007 8:49 pm
Location: Daytona Beach Shores, Florida
Contact:

Re: Goyal and Welch Paper

Post by Mel Lindauer » Wed Jun 22, 2011 8:18 am

jln wrote:This topic has been studied before, notably in a 2004 academic paper by Amit Goyal of Emory University and Ivo Welch of Yale University. Here's a link to a PDF version of their paper:

A Comprehensive Look at the Empirical Performance of Equity Premium Prediction

Here's the abstract:
Given the historically high equity premium, is it now a good time to invest in the stock market? Economists have suggested a whole range of variables that investors could or should use to predict: dividend price ratios, dividend yields, earnings-price ratios, dividend payout ratios, net issuing ratios, book-market ratios, interest rates (in various guises), and consumption- based macroeconomic ratios (cay). The typical paper reports that the variable predicted well in an in-sample regression, implying forecasting ability.
Our paper explores the out-of-sample performance of these variables, and finds that not a single one would have helped a real-world investor outpredicting the then-prevailing historical equity premium mean. Most would have outright hurt. Therefore, we find that, for all practical purposes, the equity premium has not been predictable, and any belief about whether the stock market is now too high or too low has to be based on theoretical prior, not on the empirically variables we have explored.
Among many other variables and variations of various strategies, they investigated PE10, which is the strategy being discussed in this thread. They report their results in section 4.3 on page 18:
Lamont (1998) explores variations of the E/P ratio and the payout ratio. Table 8 thus explores variations on the computation of earnings and dividend ratios. For example, Earning(10Y) are the moving average 10-year earnings. We explore two different horizons: one in which the forecast begins in 1902, another in which the forecast begins in 1964.

Panel A shows that there is both statistically significant outperformance (both in-sample and out-of-sample!) the price variables (e/p and d/p) if we use longer-term moving average price ratios and if we begin our forecasts in 1902. The economic significance reached 33 basis points for the Earnings(10Y)/Price ratio, though it is below 10 basis points for all other variations. The payout ratio (d/e) does not work. Unfortunately, if we begin our forecasts in 1964, it is not just that our variables are no longer statistically significant, they outright underperform.

Panels B through D explore 3-year, 5-year, and 10-year horizons, respectively. Panel B shows that the 3-year horizon predictions look like the one-year horizons: significant outperformance of long-memory price ratios if we begin in 1902, but underperformance if we begin in 1964. Panel C and D show that the 5-year and 10-year horizon predictions become progressively worse. The 10-year horizon predictions, however, show statistically insignificant overperformance when we begin predictions in 1964, but none if we begin predictions in 1902.

In sum, there is a tiny hint that long-memory earnings-price ratios might have better in-sample and out-of-sample performance than the prevailing mean; but the empirical evidence is so modest that it is better interpreted as not speaking against e/p, instead of speaking in favor for e/p. It looks decent primarily because the other predictive variables look so incredibly bad.
Note that the "earnings-price" ratio (EP) used by Goyal and Welch is simply the inverse of the "price-to-earnings" ratio (PE). So the results they found for EP10 apply equally well to PE10. See the paper for the tables and charts referenced in the quote above, as well as for details on methodology, other results, etc.

So Goyal and Welch are not as enthusiastic about using PE10 as are the authors of the paper discussed here. Make of this what you will. I believe that it remains a rather contentious topic in the academic world, with respected researchers on both sides of the debate. I'm not competent to judge all of this, but it is rather interesting, and if anyone is still following this long thread, I thought you might like to see a contrary point of view.

John Norstad
Nice to see your name on a post, John. Hopefully you'll stick around and enrich this forum with your vast knowledge.
Best Regards - Mel | | Semper Fi

User avatar
fredflinstone
Posts: 2101
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone » Wed Jun 22, 2011 8:26 am

tadamsmar wrote:
fredflinstone wrote:
tadamsmar wrote:
fredflinstone wrote:
Rodc wrote: It should matter to you.

People at that time priced stocks (as best they could) to give a decent return for the risk. Returns net of taxes, trading costs, etc.

When tax and other drag are high, stocks are priced to give a higher expected return to get an ok net return (for the risk).

So if you only look at returns without considering past drag you have very unrealistic values. In particular returns you can't expect to get now unless the drag is the same.

Since the drag changes over time, this is just another reason why Past Returns do not Predict Future Returns very well.

This is another one of those cases where you can't just use numbers blindly without really understanding their limitations.
That is an excellent point. I hadn't considered it. Thank you. I retract my statement that I don't care about historical tax rates. It would be good to see Pfau's analysis both ways--once including analysis of taxes and once without.

Because (a) the over-performance of the market-timing approach is large and (b) the market-timing approach does not require much trading, I am confident that market-timing will prevail over standard rebalancing in both cases. But it would be interesting to see how much the margin decreases when taxes are taken into account.
Unlike you, I agree with Pfau that more research is needed.
?? I just said more research would be "interesting" and "good." To assist the reader, I have highlighted these statements in bold and red above. It is annoying when people deliberately mischaracterize what I write. If you want to criticize my arguments, fine, but please don't criticize things I didn't say.
You said: I am confident that market-timing will prevail over standard rebalancing in both cases.

By "prevail", I guess you just mean show returns in a backtest. Or do you mean make you richer than than the Boglehead approach?
I mean I believe that market-timing will provide higher risk-adjusted returns in the backtests. That is not the same thing as saying that "more research isn't needed." I welcome additional research to fill the gaps in Pfau's paper.

Some of you do not seem to understand that all social science research has limitations. To suggest that a paper with limitations is worthless is to suggest that virtually all social science research (and probably a good deal of hard science research too) has no value. Professor Pfau's paper is important and useful despite its limitations.

User avatar
tadamsmar
Posts: 7725
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar » Wed Jun 22, 2011 8:41 am

fredflinstone wrote:
tadamsmar wrote:
fredflinstone wrote:
tadamsmar wrote:
fredflinstone wrote: That is an excellent point. I hadn't considered it. Thank you. I retract my statement that I don't care about historical tax rates. It would be good to see Pfau's analysis both ways--once including analysis of taxes and once without.

Because (a) the over-performance of the market-timing approach is large and (b) the market-timing approach does not require much trading, I am confident that market-timing will prevail over standard rebalancing in both cases. But it would be interesting to see how much the margin decreases when taxes are taken into account.
Unlike you, I agree with Pfau that more research is needed.
?? I just said more research would be "interesting" and "good." To assist the reader, I have highlighted these statements in bold and red above. It is annoying when people deliberately mischaracterize what I write. If you want to criticize my arguments, fine, but please don't criticize things I didn't say.
You said: I am confident that market-timing will prevail over standard rebalancing in both cases.

By "prevail", I guess you just mean show returns in a backtest. Or do you mean make you richer than than the Boglehead approach?
I mean I believe that market-timing will provide higher risk-adjusted returns in the backtests. That is not the same thing as saying that "more research isn't needed." I welcome additional research to fill the gaps in Pfau's paper.

Some of you do not seem to understand that all social science research has limitations. To suggest that a paper with limitations is worthless is to suggest that virtually all social science research (and probably a good deal of hard science research too) has no value. Professor Pfau's paper is important and useful despite its limitations.
I did not say the paper was worthless. I am saying that adopting an investment strategy based on an unpublished draft of a paper which, even if published, only justifies more research is a bad idea.

I agree that social science research has limitations. But leaving the taxes out of a backtest just plain lazy, in my opinion.

This is by no means the first thread here based an unpublished paper that advocates buying into and out of equities based on something or other (PE10, moving averages, your name it) where the author did not even bother to look up the historical tax laws and incorporate them into his backtest of his beloved strategy. I am sick of it.

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

Re: Goyal and Welch Paper

Post by Rodc » Wed Jun 22, 2011 8:42 am

Mel Lindauer wrote:
jln wrote:This topic has been studied before, notably in a 2004 academic paper by Amit Goyal of Emory University and Ivo Welch of Yale University. Here's a link to a PDF version of their paper:

A Comprehensive Look at the Empirical Performance of Equity Premium Prediction

Here's the abstract:
Given the historically high equity premium, is it now a good time to invest in the stock market? Economists have suggested a whole range of variables that investors could or should use to predict: dividend price ratios, dividend yields, earnings-price ratios, dividend payout ratios, net issuing ratios, book-market ratios, interest rates (in various guises), and consumption- based macroeconomic ratios (cay). The typical paper reports that the variable predicted well in an in-sample regression, implying forecasting ability.
Our paper explores the out-of-sample performance of these variables, and finds that not a single one would have helped a real-world investor outpredicting the then-prevailing historical equity premium mean. Most would have outright hurt. Therefore, we find that, for all practical purposes, the equity premium has not been predictable, and any belief about whether the stock market is now too high or too low has to be based on theoretical prior, not on the empirically variables we have explored.
Among many other variables and variations of various strategies, they investigated PE10, which is the strategy being discussed in this thread. They report their results in section 4.3 on page 18:
Lamont (1998) explores variations of the E/P ratio and the payout ratio. Table 8 thus explores variations on the computation of earnings and dividend ratios. For example, Earning(10Y) are the moving average 10-year earnings. We explore two different horizons: one in which the forecast begins in 1902, another in which the forecast begins in 1964.

Panel A shows that there is both statistically significant outperformance (both in-sample and out-of-sample!) the price variables (e/p and d/p) if we use longer-term moving average price ratios and if we begin our forecasts in 1902. The economic significance reached 33 basis points for the Earnings(10Y)/Price ratio, though it is below 10 basis points for all other variations. The payout ratio (d/e) does not work. Unfortunately, if we begin our forecasts in 1964, it is not just that our variables are no longer statistically significant, they outright underperform.

Panels B through D explore 3-year, 5-year, and 10-year horizons, respectively. Panel B shows that the 3-year horizon predictions look like the one-year horizons: significant outperformance of long-memory price ratios if we begin in 1902, but underperformance if we begin in 1964. Panel C and D show that the 5-year and 10-year horizon predictions become progressively worse. The 10-year horizon predictions, however, show statistically insignificant overperformance when we begin predictions in 1964, but none if we begin predictions in 1902.

In sum, there is a tiny hint that long-memory earnings-price ratios might have better in-sample and out-of-sample performance than the prevailing mean; but the empirical evidence is so modest that it is better interpreted as not speaking against e/p, instead of speaking in favor for e/p. It looks decent primarily because the other predictive variables look so incredibly bad.
Note that the "earnings-price" ratio (EP) used by Goyal and Welch is simply the inverse of the "price-to-earnings" ratio (PE). So the results they found for EP10 apply equally well to PE10. See the paper for the tables and charts referenced in the quote above, as well as for details on methodology, other results, etc.

So Goyal and Welch are not as enthusiastic about using PE10 as are the authors of the paper discussed here. Make of this what you will. I believe that it remains a rather contentious topic in the academic world, with respected researchers on both sides of the debate. I'm not competent to judge all of this, but it is rather interesting, and if anyone is still following this long thread, I thought you might like to see a contrary point of view.

John Norstad
Nice to see your name on a post, John. Hopefully you'll stick around and enrich this forum with your vast knowledge.
Yes. Indeed I missed this excellent post somehow on first reading this page.

This is perhaps the best post in all 7 pages.

So thanks John.
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.

chaster86
Posts: 194
Joined: Mon May 16, 2011 12:15 pm

Post by chaster86 » Wed Jun 22, 2011 10:36 am

fredflinstone wrote: I can't speak for Professor Pfau, but I care not one whit about historical capital gains rates. The tax rate in 1898 or 1921 or whatever has no bearing on my investment strategy. All that matters to me are current and future tax rates.
You're missing tadamsmar's point.

His point is that it may be the only reason it appears your market timing strategy would have had such large gains in the past is because high capital gains taxes prevented it from being profitable which is what caused the market participants of the time to leave all that money on the table. So if current or future capital gains rates are lower than historical capital gains rates, then the market timing approach won't have large gains anymore because the gains will be spread out among millions of market participants who suddenly have incentive to implement such strategies (and if those participants adopt a "stay the course" philosophy then those gains could even turn into loses..."stay the course" only works with the market portfolio because the market portfolio can never lose).

Beating the market is like hunting endangered species; It seems like it will work when it's illegal or hidden (ie highly taxed or no evidence of a clear advantage), but once it becomes legal or known (ie lower current/future tax rates or professor Pfau publishes an irrefutable study) the flood of hunting makes the species go extinct.

So the tax rate in 1898 or 1921 has bearing on your investment strategy because it is very likely the cause of the market anomaly you're trying to exploit. If current and future tax rates are different from historical rates, then the market will also probably behave differently and your investment strategy won't work.

Is that clear?
fredflinstone wrote: Not only that, but some of the above (plus Bogle) advocate various forms of market-timing with respect to bonds or TIPS. They may not call it market-timing, but in my view, that is what it is.
Has anyone ever said that John Bogle is infallible? Yes, Bogle advocates a little market timing. Does that mean it's a good idea to take a minor flaw and expand it into a major flaw? No.

Your argument is akin to saying that since the president admitted to smoking marijuana, you're going to start giving your children cocaine for breakfast. It's backwards.

Flaws should be corrected or (if there's no clearly advantageous correction available) left alone, not compounded.
fredflinstone wrote:Ignoring valuations seems to me to be foolish.
Why would it be foolish? The value of an investment is only what someone else will pay you for it. Is there any reason to believe that company earnings are a better predictor of this than market capitalization? Perhaps political sway, or information, or any number of other assets will be more valuable in the coming years.

Those that study strategies for the "buy and sell game", find that valuations can't help a winning strategy because, since we always buy and sell at the market price, it is the valuations of others which determine our success or failure. Since we can't read the minds of the other market participants (who would certainly like to fool us and take our money), to me it seems that ignoring what they say (ie valuations) and watching what they do (ie market cap) would be the wisest approach.

letsgobobby
Posts: 11421
Joined: Fri Sep 18, 2009 1:10 am

Post by letsgobobby » Wed Jun 22, 2011 11:17 am

Those arguing to ignore PE10 would want to hold the exact same stock/bond allocation in 2000 (PE10 41) as 2009 (PE10 11)? Maybe that is the reason, prima facie, for the persistence of PE10 as a useful marker!

User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 8:51 pm

Post by bob90245 » Wed Jun 22, 2011 11:47 am

letsgobobby wrote:Those arguing to ignore PE10 would want to hold the exact same stock/bond allocation in 2000 (PE10 41) as 2009 (PE10 11)? Maybe that is the reason, prima facie, for the persistence of PE10 as a useful marker!
An alternative reason to ignore P/E10 is because some hold portfolios different than S&P 500 and bonds at only 100/0 or 0/100. One example:

Image
Source: http://www.coffeehouseinvestor.com/the- ... e-returns/

“Build wealth, ignore Wall Street and get on with your life.”
-- Bill Schultheis
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

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

Post by Rodc » Wed Jun 22, 2011 11:54 am

letsgobobby wrote:Those arguing to ignore PE10 would want to hold the exact same stock/bond allocation in 2000 (PE10 41) as 2009 (PE10 11)? Maybe that is the reason, prima facie, for the persistence of PE10 as a useful marker!
Might be useful to read John Norstad's post above before stating that that PE10 has had persistent usefulness.

That said, at truly extreme values it may have value. In the very broad gray middle, not so much (if at all).
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.

SP-diceman
Posts: 3968
Joined: Sun Oct 05, 2008 9:17 am

Post by SP-diceman » Wed Jun 22, 2011 12:02 pm

So the tax rate in 1898 or 1921 has bearing on your investment strategy
I doubt they were high in 1898. :)

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Post by lazyday » Wed Jun 22, 2011 12:03 pm

fredflinstone wrote:Professor Pfau's paper is important and useful despite its limitations.
I don't contest the value of his work. But do you have the expertise and experience to determine that his paper is "important and useful"? To decide that it is useful enough to act on?
fredflinstone wrote:The position I am advocating is not identical to the one Pfau modeled in his paper. Rather, my position is basically the same as that advocated by Jack Bogle: buy-hold-and-rebalance for most of the portfolio but allow some "gentle" market timing at the margins when valuations are extreme.
Some days or weeks ago, you posted specific tentative plans I considered aggressive and not similar to the advice of Bogle. If you come up with something more conservative, that either is in line with advice of Bogle, or waits for truly extreme valuations to make large changes (my own diversion from Bogle's advice) then I might not find fault in your plans, other than maybe execution risk, including perhaps what led you to adopt the rules.

If you have some history of buying/rebalancing at market lows, and selling high, and have belief in the plans other than just from one study (since there's too much risk other studies will counter it, even perhaps form the same author!) then I might think your plans are better than buy&hold&rebalance, for those who can stick to them.

DP
Posts: 659
Joined: Thu Apr 17, 2008 5:19 pm

Post by DP » Wed Jun 22, 2011 12:03 pm

Hi,
In fact I have some hope that enough people now use P/E10 and other valuation metrics that it helps damp down highs and lows and could be part of why we never hit single digit lows in the last crash (but we likely will never know).
I've stayed quiet in this discussion as I see no need to involve myself in what I see as valid arguments for and against this strategy, but the above comment jumped out at me. There are not many things that I think I know for sure, but one is that human nature is not likely to change to any great extent. There is absolutely no reason that human history of irrational crowd behavior is suddenly a thing of the past. It WILL be repeated in the future. That the market was not driven to single digits lows was not because humans are suddenly more rational so it had to be other reasons (cheap money?).

Don

SP-diceman
Posts: 3968
Joined: Sun Oct 05, 2008 9:17 am

Post by SP-diceman » Wed Jun 22, 2011 12:12 pm

DP wrote:Hi,
In fact I have some hope that enough people now use P/E10 and other valuation metrics that it helps damp down highs and lows and could be part of why we never hit single digit lows in the last crash (but we likely will never know).
I've stayed quiet in this discussion as I see no need to involve myself in what I see as valid arguments for and against this strategy, but the above comment jumped out at me. There are not many things that I think I know for sure, but one is that human nature is not likely to change to any great extent. There is absolutely no reason that human history of irrational crowd behavior is suddenly a thing of the past. It WILL be repeated in the future. That the market was not driven to single digits lows was not because humans are suddenly more rational so it had to be other reasons (cheap money?).

Don
Excellent point.


Thanks
SP-diceman

chaster86
Posts: 194
Joined: Mon May 16, 2011 12:15 pm

Post by chaster86 » Wed Jun 22, 2011 4:18 pm

letsgobobby wrote:Those arguing to ignore PE10 would want to hold the exact same stock/bond allocation in 2000 (PE10 41) as 2009 (PE10 11)?
Ex-post or ex-ante?

There's going to be a big change as the world economy shifts to an information economy. It has to happen because continued exponential growth of physical goods is simply physically impossible.

You can ridicule the dot com bubble all you want, but if the world keeps progressing the way it has then those stock prices will prove a lot more realistic than your PE10 valuations. Do you want to miss out when the time finally comes? Third time's a charm.

jln
Posts: 178
Joined: Thu Dec 09, 2010 2:18 pm

Paul Samuelson on Timing and Risk

Post by jln » Wed Jun 22, 2011 4:45 pm

Phau says the following in his paper on page 4:
Ex-ante, the market-timing strategies have an average asset allocation of 50 percent stocks. For a wide variety of risk measures, the 100 percent stocks strategy is noticeably more risky than market timing, while an alternative 50/50 fixed allocation strategy provides broadly comparable risks as market timing.
This is a common error. An allocation that is 0% stocks half the time and 100% stocks half the time is actually considerably more risky than a fixed 50/50 allocation over the full time period. The risks are not "broadly comparable." (In this context "risk" means the usual volatility as measured by variance or standard deviation.)

I believe the great economist Paul Samuelson was the first to notice this. He summarizes this nicely in his 1994 paper The Long-Term Case for Equities - And how it can be oversold" in The Journal of Portfolio Management:
As revealed in this Journal (Samuleson [1990]), it is a fatal mathematical fallacy to think that diversifying across time - as, for example, when you hold zero-one hundred stocks-bonds a third of the time and one hundred-zero two-thirds of the time - it just as variance-reducing as to hold sixty-six and two-thirds-thirty-three and one-third all the time. You pay through the nose in excess volatility from diversifying primarily across time rather than at every period of time. Only geniuses at timing or lucky sailors and drunks can be ahead when paying such a secret toll.
The 1990 paper Samuelson mentions in this quote is titled Asset allocation could be dangerous to your health. ("Asset allocation" here means what today we call "market timing".) He works out all the math in that paper. Here's just a few quotes from the non-math parts of that paper:
... the case of an MIT professor who in one year went back and forth sixteen times between all-equities and zero-equities. ... because no academic fool could be so unlucky as to have been wrong every time in his moves, one at first careless thought might suspect that he had done himself little harm. While his complacent colleagues stayed all year long half-and-half in stocks and fixed-principal assets, this professor would seem to have achieved his half-and-half mix by across-time diversification (so to speak, Monday-Wednesday-Friday in stocks and Tuesday-Thursday-Saturday in bonds). ... this evaluation would indeed be a careless underestimation of the inefficiency of across-time diversification.

... I must not pretend to prove too much. It you do have timing ability, flaunt it! But in the absence of Napoleonic pretensions to clairvoyance, your rational flauntings are more likely to involve switches of a few percent in your equity fraction around some optimal intermediate level that the swings from 100% in stocks to 10% in stocks that characterized many asset allocation systems in the last two years.
(I quote these passages above to make a point, but I confess also because I love the way Samuelson writes and I think he's very funny.)

What does this observation of Samuelson's mean in terms of the PE10 timing scheme in the Phau paper?

What Samuelson shows us is that we are starting in a hole. Our 0% stocks half the time and 100% stocks half the time strategy is actually considerably riskier (more volatile) than a 50/50 fixed portfolio if we assume returns are random. Thus our PE10 timing strategy must be able to fairly reliably forecast the following year's return just to break even with the 50/50 buy-and-hold strategy (on a risk-adjusted basis).

Couple this observation with the very weak results for PE10 forecasting ability that Goyal and Welch found in the paper I discussed earlier.

I don't claim to be refuting the Phau paper at all. Again, I'm not competent to judge all of this (it's "above my pay grade", as a professor I work with likes to say). But it does cast some doubt, I think, and it certainly worth thinking about in any case.

Here's one more Samuelson quote that's on topic but is mostly just for fun:
A notorious temptation is to try to "time the general market." ... Sell just before or just after the whole market peaks. Park your profits in money market funds. Then, just before or just after the whole market bottoms out, buy back into it. Your profits in comparison with buy-and-hold will be enormous. If you can do it. If. Experience shows that out of a thousand money managers and private investors only a small fraction in fact succeed in buying back at a lower average price than the have sold out at.

That's the mediocre crowd. You are wonderful you. But, as Thomas Cromwell said: "I beseech you in the bowels of Christ to consider that (wonderful) you may be wrong." Or may be sometimes wrong. Among the asset allocators there are confident ones; they move from ninety-ten in stocks-bonds to five-ninety-five in stocks-bonds. That implies a degree of self-confidence bordering on hubris and self-deception.
John Norstad

letsgobobby
Posts: 11421
Joined: Fri Sep 18, 2009 1:10 am

Post by letsgobobby » Wed Jun 22, 2011 5:30 pm

PE10 works best for 10-20 year intervals... not 12 month or 7 years.

I don't think anyone would responsibly advocate swinging between 0% and 100% anything. What about between 20% stocks and 80% stocks?

I am not an economist or a mathemitican and I acknowledge many Bogleheads' superiority in those fields. Indulge me with a logical rather than numerical question.

Given "E", PE10 is entirely psychology, because "P" for that E is completely dependent on the mood of investors at that time. Warren Buffett has said it is unrealistic to expect corporate earnings to grow much faster than GDP over any long period of time. And GDP can vary but typically doesn't, at least not much, over very long periods of time. Therefore if psychology is constant, PE10 should be constant; if psychology is not contant, PE10 will vary.

Maybe i've read too much Kindleberger, but it seems to me that human nature regarding speculative booms and busts has been remarkably constant in it cyclicity. And that it would be foolhardy to expect that to change.

How am I misunderstanding this?

User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 8:51 pm

Post by bob90245 » Wed Jun 22, 2011 5:45 pm

letsgobobby wrote:Given "E", PE10 is entirely psychology, because "P" for that E is completely dependent on the mood of investors at that time. Warren Buffett has said it is unrealistic to expect corporate earnings to grow much faster than GDP over any long period of time. And GDP can vary but typically doesn't, at least not much, over very long periods of time. Therefore if psychology is constant, PE10 should be constant; if psychology is not contant, PE10 will vary.

Maybe i've read too much Kindleberger, but it seems to me that human nature regarding speculative booms and busts has been remarkably constant in it cyclicity. And that it would be foolhardy to expect that to change.

How am I misunderstanding this?
Real E1 and real E10 vary; the former, a lot more.

Image
Source: http://bobsfinancialwebsite.com/OddsAndEnds.html

Would be interesting to see a chart of both real E10 and real GDP.
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

Yipee-Ki-O
Posts: 167
Joined: Sun May 24, 2009 12:42 pm

Post by Yipee-Ki-O » Wed Jun 22, 2011 6:17 pm

bob90245 wrote:
letsgobobby wrote:Those arguing to ignore PE10 would want to hold the exact same stock/bond allocation in 2000 (PE10 41) as 2009 (PE10 11)? Maybe that is the reason, prima facie, for the persistence of PE10 as a useful marker!
An alternative reason to ignore P/E10 is because some hold portfolios different than S&P 500 and bonds at only 100/0 or 0/100. One example:

Image
Source: http://www.coffeehouseinvestor.com/the- ... e-returns/

“Build wealth, ignore Wall Street and get on with your life.”
-- Bill Schultheis
Exactly. If one is only interested in holding the S&P 500 Index (or its nearly identical twin, the Total Stock Market Index Fund), then this discussion about the merits of an S&P 500 Index buy-hold-rebalance portfolio versus timing the S&P 500 is most useful. If one is a "splitter" and holds several different asset classes, not so much. But I enjoy the discussion nonetheless!

letsgobobby
Posts: 11421
Joined: Fri Sep 18, 2009 1:10 am

Post by letsgobobby » Wed Jun 22, 2011 6:34 pm

obviously S&P500 PE10 can only be used for the S&P500 portion of one's portfolio. For a 2 fund portfolio that could easily be 50% of the total. In the example Bob gave, it's roughly 20% and would clearly be less meaningful.

SP-diceman
Posts: 3968
Joined: Sun Oct 05, 2008 9:17 am

Post by SP-diceman » Wed Jun 22, 2011 7:03 pm

letsgobobby wrote: I am not an economist or a mathemitican and I acknowledge many Bogleheads' superiority in those fields. Indulge me with a logical rather than numerical question.
To be honest I think math/economic skills are irrelevant.
If this was important, mathematicians and economists would be millionaires in the market.
(you see I can make up the same accusations)

All I’ve seen so far is “ I disagree”

No one has proven anything.

Its simple to point out danger.
I’m sure everyone here who disagrees with PE10 has driven in a car.
Ever see what happens when a car hit’s a tree at 60 mph?
Cars today drive faster and are made of cheaper materials.
Who knew Bogleheads were so risky?

Of course I cant prove you will crash.
All I can do is point out crashes exist. (and that today is different)

Have your odds changed any because I pointed it out????



Thanks
SP-diceman

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

Post by Rodc » Wed Jun 22, 2011 9:22 pm

DP wrote:Hi,
In fact I have some hope that enough people now use P/E10 and other valuation metrics that it helps damp down highs and lows and could be part of why we never hit single digit lows in the last crash (but we likely will never know).
I've stayed quiet in this discussion as I see no need to involve myself in what I see as valid arguments for and against this strategy, but the above comment jumped out at me. There are not many things that I think I know for sure, but one is that human nature is not likely to change to any great extent. There is absolutely no reason that human history of irrational crowd behavior is suddenly a thing of the past. It WILL be repeated in the future. That the market was not driven to single digits lows was not because humans are suddenly more rational so it had to be other reasons (cheap money?).

Don
People change behavior to changes in their environment all the time. Change tax laws they adapt. Given then new tools, they use them to do new stuff. Arbitrage frequently take away any such advantage, one of the more sure things we know about investing. Does that apply to P/E10? Maybe or maybe not.

The good news is I'm smart enough to know I'm not smart enough to make money by trying to market time or otherwise do stupid stuff with such idle musings. :)
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

Post by Rodc » Wed Jun 22, 2011 9:43 pm

All I’ve seen so far is “ I disagree”
Then you missed John Norstad's post and link.

Proof? No, there is no proof to any of this. But certainly more than simply "I disagree".
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.

Yipee-Ki-O
Posts: 167
Joined: Sun May 24, 2009 12:42 pm

Post by Yipee-Ki-O » Wed Jun 22, 2011 9:46 pm

letsgobobby wrote:obviously S&P500 PE10 can only be used for the S&P500 portion of one's portfolio. For a 2 fund portfolio that could easily be 50% of the total. In the example Bob gave, it's roughly 20% and would clearly be less meaningful.
And according to Wade Pfau in his study, Revisiting the Fisher and Statman Study, the study referenced by the OP, "[t]he buy-and-hold strategy is represented by 100 percent large-capitalization stocks (S&P 500)". I merely point out, as did bob90245, that for the purposes of this study which compares returns for a buy-hold-rebalance strategy versus a market-timing approach, stocks equal the S&P 500 Index. Obviously for anyone who holds other equity class assets other than the S&P 500 Index, the findings of the study and the discussion of those findings are indeed clearly less meaningful.

User avatar
Dick Purcell
Posts: 520
Joined: Tue Oct 26, 2010 1:58 am

Post by Dick Purcell » Thu Jun 23, 2011 12:56 am

Wade –

I’d like to suggest a direction of further research in your PE10-based investing project that I think has very good chances for delivering real practical "I-can-use-it-now" value for the individual investor.

Instead of using PE10, find or originate a new index I’ll call PR10f. The R10 means annual arithmetic mean for 10 years of returns. The f means future returns, for the 10 years following after the P.

(Maybe 10 is not the best number of years. Might test other numbers.)

I envision a slope, of the form R = Rbase – ½ * (P – Pbase)

In words, my equation is intended to say that for every 1 above (below) average for P, R estimated for next 10 years is “expected” to be ½ % lower (higher) than otherwise projected. Of course, this is just a preliminary suggestion for the form of the suggested-research result. The ½, which I have drawn from thin air, is to be replaced by something with evidence.

I think this approach offers the following advantages:

>> Each investor can apply the adjustment to projected “expected” R in whatever projecting he does for his own $$ plan and goals, straight-line or Monte or whatever.

>> If used in Monte, the investor can adjust the “expected” rate while continuing to incorporate standard deviation, so what is done is slanting prospects up or down a little while continuing to incorporate those big random uncertainties.

>> I think investors will commonly find that the implications for their particular plans are not only about allocation, but also outlook for future returns and prudent adjustments of cash flow plans. I expect that commonly the implications for future return and cash flow adjustment will greatly outweigh any indications that allocations should be changed.

>> In implications for allocations, I think that for the individual investor’s plan and goals, whatever allocation adjustments are suggested by the analysis for that individual’s particular plan and goals will sometimes be different from the abstract generalities commonly cited in this thread. For example, Bongleur suggested earlier in this thread that for a particular investor, a low PE10 might suggest reducing the stocks allocation, because with improved R outlook for stocks the investor can maintain high probability of meeting his goals with less of stocks uncertainty.

A major reason for my high hopes that this approach will appear to have value is my expectation for significant implications for future return and prudent cash flow adjustment, rather than for allocations. I base this expectation partly on your other research regarding SWR and Safe Saving Rate, where your data suggest very dramatic implications of retirement-day PE10 for following years' SWR.

Most important, I think this approach can advance your research on PE10-based investing from “academic” to “Hey this is in a form that I the individual investor can pop right into my own projections to see implcations for my particular $$ plan and goals, today!” In other words, what you’ve called “goals-based investing.”

I think in your other projects you have distinguished yourself from the great bulk of university investment research, publications, and teaching in producing results conceived and designed for people to understand and use. I think with the approach suggested above, you can do the same with your project on PE10-based investing.

Dick Purcell

User avatar
tadamsmar
Posts: 7725
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar » Thu Jun 23, 2011 3:04 am

letsgobobby wrote:PE10 works best for 10-20 year intervals... not 12 month or 7 years.

I don't think anyone would responsibly advocate swinging between 0% and 100% anything. What about between 20% stocks and 80% stocks?

I am not an economist or a mathemitican and I acknowledge many Bogleheads' superiority in those fields. Indulge me with a logical rather than numerical question.

Given "E", PE10 is entirely psychology, because "P" for that E is completely dependent on the mood of investors at that time. Warren Buffett has said it is unrealistic to expect corporate earnings to grow much faster than GDP over any long period of time. And GDP can vary but typically doesn't, at least not much, over very long periods of time. Therefore if psychology is constant, PE10 should be constant; if psychology is not contant, PE10 will vary.

Maybe i've read too much Kindleberger, but it seems to me that human nature regarding speculative booms and busts has been remarkably constant in it cyclicity. And that it would be foolhardy to expect that to change.

How am I misunderstanding this?
You are misunderstanding this by assuming that all estimates of earnings growth are due to "psychology".

User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade » Thu Jun 23, 2011 6:17 am

Dick Purcell wrote:Wade –

I’d like to suggest a direction of further research in your PE10-based investing project that I think has very good chances for delivering real practical "I-can-use-it-now" value for the individual investor.

Instead of using PE10, find or originate a new index I’ll call PR10f. The R10 means annual arithmetic mean for 10 years of returns. The f means future returns, for the 10 years following after the P.

(Maybe 10 is not the best number of years. Might test other numbers.)

I envision a slope, of the form R = Rbase – ½ * (P – Pbase)

In words, my equation is intended to say that for every 1 above (below) average for P, R estimated for next 10 years is “expected” to be ½ % lower (higher) than otherwise projected. Of course, this is just a preliminary suggestion for the form of the suggested-research result. The ½, which I have drawn from thin air, is to be replaced by something with evidence.

I think this approach offers the following advantages:

>> Each investor can apply the adjustment to projected “expected” R in whatever projecting he does for his own $$ plan and goals, straight-line or Monte or whatever.

>> If used in Monte, the investor can adjust the “expected” rate while continuing to incorporate standard deviation, so what is done is slanting prospects up or down a little while continuing to incorporate those big random uncertainties.

>> I think investors will commonly find that the implications for their particular plans are not only about allocation, but also outlook for future returns and prudent adjustments of cash flow plans. I expect that commonly the implications for future return and cash flow adjustment will greatly outweigh any indications that allocations should be changed.

>> In implications for allocations, I think that for the individual investor’s plan and goals, whatever allocation adjustments are suggested by the analysis for that individual’s particular plan and goals will sometimes be different from the abstract generalities commonly cited in this thread. For example, Bongleur suggested earlier in this thread that for a particular investor, a low PE10 might suggest reducing the stocks allocation, because with improved R outlook for stocks the investor can maintain high probability of meeting his goals with less of stocks uncertainty.

A major reason for my high hopes that this approach will appear to have value is my expectation for significant implications for future return and prudent cash flow adjustment, rather than for allocations. I base this expectation partly on your other research regarding SWR and Safe Saving Rate, where your data suggest very dramatic implications of retirement-day PE10 for following years' SWR.

Most important, I think this approach can advance your research on PE10-based investing from “academic” to “Hey this is in a form that I the individual investor can pop right into my own projections to see implcations for my particular $$ plan and goals, today!” In other words, what you’ve called “goals-based investing.”

I think in your other projects you have distinguished yourself from the great bulk of university investment research, publications, and teaching in producing results conceived and designed for people to understand and use. I think with the approach suggested above, you can do the same with your project on PE10-based investing.

Dick Purcell
There are lots of interesting discussions in this thread, and I do need to go back through and read it very carefully.

While most everyone agrees that valuations matter, we have strong disagreement about whether investors can take advantage of this information. At this point I still lean toward the conclusion that investors with a long-term horizon can benefit from considering valuations in their asset allocation, but I do understand the reasonable case which is being made that it is better to just leave your asset allocation alone in this regard.

But as Dick is saying here, it is definitely the case that investors may be able to consider valuations in ways other than changing their asset allocation. Dick is even suggesting that these other considerations are probably more important than the asset allocation angle. That's a good question.

As I've been saying before, the study discussed here was only meant to be a secondary study to be finished after a primary study about individual investors. But that primary study is no where near complete and only exists as a blog entry now, and frankly there are lots of other interesting issues I want to look at which are pushing ahead of getting that study finished anytime soon.

Nonetheless, one thing that the primary study is missing is any type of investor goals, except broadly how to increase risk-adjusted returns for your retirement savings. Some more specific goals are needed.

At the same time, my paper about "safe savings rates" is just as dependent on PE10 as is the paper here about the Fisher and Statman study. But for "safe savings rates", I use a fixed asset allocation and that paper does not generate anywhere near as much controversy. But valuations are very important in that paper, as this is the reason for the baseline individual that their SAFEMIN safe savings rate to meet their retirement goals is only 16.6% instead of 37.7%.

I'm still trying to get my head wrapped around all this issues. As the stock market goes up, you are hopefully accumulating wealth more quickly which should be getting you closer to your goals and reducing your need for risk, while at the same time valuations are probably increasing as well, which does connect (in a controversial way) to lower expected future returns. Also, as time is passing, the time horizon for an individual is shortening. Bongleur's point is also quite interesting.

Anyway, I don't really have a main point with this post, but just to thank Dick for his very enlightening post, as well as all the discussants here.

cjking
Posts: 1797
Joined: Mon Jun 30, 2008 4:30 am

Post by cjking » Thu Jun 23, 2011 7:25 am

rrosenkoetter wrote:
fredflinstone wrote:
Mel Lindauer wrote:It appears from this thread that at least a few folks think this is the Holy Grail of investing, and leaving this kind of stuff unchallenged is dangerous.
If you ask me, a few folks think the standard Boglehead approach (fixed asset allocation; stay the course regardless of valuation) is the Holy Grail of Investing.
If you stay the course, you will get market returns. There's no magic there.

If you market time based on valuation or whatever, you may get more than market returns or less than market returns. The odds for each possibility vary per method used.

For my retirement money, I will take the market returns. I'm already gambling because no one knows what those returns will be in the future... we assume they will general be upwards over long periods of time...

I'm not going to add yet another gamble on top of that hoping to make more than market returns.
If timing has no merit then its effect should be broadly neutral, so to "add yet another gamble on top of that" would at worst make no difference to your risk-adjusted returns.

If you choose a "timing" strategy with an eye to what could go wrong, you have a sort of inverse Pascal's wager. You're not worse off if the strategy has no merit, but you benefit if it does.

cjking
Posts: 1797
Joined: Mon Jun 30, 2008 4:30 am

Post by cjking » Thu Jun 23, 2011 7:39 am

DP wrote:Hi,
In fact I have some hope that enough people now use P/E10 and other valuation metrics that it helps damp down highs and lows and could be part of why we never hit single digit lows in the last crash (but we likely will never know).
I've stayed quiet in this discussion as I see no need to involve myself in what I see as valid arguments for and against this strategy, but the above comment jumped out at me. There are not many things that I think I know for sure, but one is that human nature is not likely to change to any great extent. There is absolutely no reason that human history of irrational crowd behavior is suddenly a thing of the past. It WILL be repeated in the future. That the market was not driven to single digits lows was not because humans are suddenly more rational so it had to be other reasons (cheap money?).

Don
The answer to this dispute may not matter much. It isn't necessarily a problem (to a timer) if the use of PE10 dampens oscillations in future. A reasonable strategy would degenerate into its buy-and-hold equivalent if this does happen, but yield a premium if it doesn't.

User avatar
fredflinstone
Posts: 2101
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone » Thu Jun 23, 2011 7:44 am

According to Pfau, Graham & Dodd "suggested that the medium allocation be maintained as long as PE10 falls within a band between 2/3 and 4/3 of its historical average, and only switches to more extreme allocations when PE10 moves beyond these bounds." Source: http://wpfau.blogspot.com/2011/03/long- ... s-and.html

Elsewhere, I believe Graham & Dodd have recommended always keeping between 25 percent and 75 percent of one's portfolio in equities.

In addition, I want an asset allocation strategy that takes into account both my age (lower allocation to equities as one gets older) and expected returns on bonds (so a lower allocation to equities as TIPS yields rise).

Here is an algorithm for a moderately risk-averse investor that is consistent with the above. I believe it is also completely consistent with Jack Bogle's suggestion to engage in "gentle market timing" when valuations are extreme:

If PE10 if below 11:
% in equities = INT(MAX(25,((80-AGE)+(2*(11-PE10))-(10*TIPS))))

if PE10 is between 11 and 22:
% in equities = INT(MAX(25,((80-AGE)-(10*TIPS))))

If PE10 is 23 or above:
% in equities = INT(MAX(25,((80-AGE)+(2*(23-PE10))-(10*TIPS))))

where "AGE" is the investor's age, "PE10" is the PE10 ratio, and "TIPS" is the 5-year TIPS yield.

I realize the choice of 5-year TIPS yields is unusual. I chose that duration because, for me, the most likely alternative to equities is intermediate-term municipal bonds. The 5-year TIPS yield is a proxy for the real return I can expect to earn on my intermediate-term muni holdings.

Of course, a more aggressive investor could modify the formulas as needed (e.g., replace "80-AGE" with "110-AGE" or what have you).

Sample values:

<img src="http://i55.tinypic.com/2ccxkxx.jpg" width="800"></img>


Thoughts?
Last edited by fredflinstone on Thu Jun 23, 2011 10:14 am, edited 1 time in total.

wellmoneyed
Posts: 267
Joined: Sun May 03, 2009 10:08 am

Post by wellmoneyed » Thu Jun 23, 2011 8:24 am

fredflinstone wrote:Here is an algorithm for a moderately risk-averse investor that is consistent with the above. I believe it is also completely consistent with Jack Bogle's suggestion to engage in "gentle market timing" when valuations are extreme:

If PE10 if below 11:
% in equities = INT(MAX(25,((80-AGE)+(2*(11-PE10))-(10*TIPS))))
......
Thoughts?
Four points:

1. I would guess Bogle would NOT be supportive of this approach.
2. I found the the earlier argument of Dogs of the Dow compelling. How is PE10 different from Dogs of the Dow 10 years ago?
3. I don't see how you can assume the effect is linear. I believe it is agreed on this board that 80/20 (bonds/stocks) is lower risk than 100 bonds. This relationship is NOT linear and likely not intuitively obvious. It appears you are assuming the PE10 alpha is linear with stock holdings and not some type of curve.
4. If you are buying stocks other than US, why do you think PE10 works in other markets/sectors?

EDIT: http://www.businessweek.com/investor/co ... page_2.htm

User avatar
DRiP Guy
Posts: 2237
Joined: Tue Feb 20, 2007 4:54 pm

Post by DRiP Guy » Thu Jun 23, 2011 8:41 am

cjking wrote:If timing has no merit then its effect should be broadly neutral..
How does this follow?

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Post by lazyday » Thu Jun 23, 2011 9:01 am

fredflinstone wrote:Here is an algorithm for a moderately risk-averse investor that is consistent with the above. I believe it is also completely consistent with Jack Bogle's suggestion to engage in "gentle market timing" when valuations are extreme:
....
Thoughts?
Looks much better than the last one. Might even be reasonable, with some caveats:

Seems overly sensitive to TIPS yield, but I'm not sure of this.

Still is probably not what Bogle meant. Wouldn't stop me from doing it, but if Bogle's opinion is an important justification to you, you might want to reread those pages from Bogle, to be sure of your own opinion of his meaning. I think he meant something much more conservative than even this version of your PE10/TIPS timing.

Execution risk. Requires dedication in panics and bubbles. Have you demonstrated ability to buy in panic and sell in bubbles before?
If you make changes to the formulas after implementation, will you only make changes that cause you to buy low and sell high? (Many people "fix" or "improve" AA and buy high or sell low.)

PE10 is only one measure of valuation, even if it might be the best single number that can be methodically calculated today. You're ignoring TTM P/E, P/B, dividend yield, P/CF, P/S, Gordon Equation, Tobin's Q, etc. If all those overwhelmingly tell another story, do you stick to your model?

PE10 is just for US TSM or US S&P 500, right? What if another asset class is cheap, such as in 1999? Some highly regarded value investors believe there is usually a cheap or reasonably priced asset class, with 2007ish the only exception. Your model will have you move into TIPS when US TSM is expensive, but US small might be cheap, or EM, or Japan, REITS, etc, based on PE10 or other measures.

Graham was probably writing about individual stocks, not the market, if that is an important justification to you. I'm not sure though, I haven't read the quote.

Post Reply