Here's a statement of the obvious: The opinions expressed here are those of the participants, not those of the Mutual Fund Observer. We cannot vouch for the accuracy or appropriateness of any of it, though we do encourage civility and good humor.
Support MFO
Donate through PayPal
The Single Greatest Predictor of Future Stock Market Returns
"The metric in this chart takes no input from any variables traditionally associated with valuation: earnings, book values, profit margins, discount rates, etc. It consists only of a simple ratio between two numbers that can easily be calculated in FRED. Yet, as a predictor of future stock market returns, it dramatically outperforms all other stock market valuation metrics commonly cited."
I like the way this guy thinks! Thanks for the link @bee!
His premise is similar to my thesis posted earlier in the context of discussing stock multiples. That stocks have become derivatives in the direction based on future investor availability than any fundamentals while we use the latter to keep a pretense that it is not.
I have not gone through his quantitative model yet to see if it makes sense and skeptical that there is predictive capability to this thesis but I am keeping an open mind.
Thanks for bringing this to the board. At first read I follow a good bit of the thinking and logic that it expresses. However, I am not so sure the markets follow logic and there is a group called the Plunge Protection Team (PPT). The PPT can have a great effect on the markets as policy changes within the FOMC itself. With this, these policies and changes thereof are directed to the Federal Reserve and its member banks. Some of the key figures that sit on the PPT come from some of the country's largest banks and also include the Treasury, the FOMC chairperson and others.
For those interested I have linked more information below on the PPT.
Perhaps, our markets are more influenced by the banks than ones wishes to believe. In any event if the markets are trading above normal p/e ratios then overall returns will be less by my thinking. This is one of the reasons I have been reducing my allocation to equities over the past couple of years as p/e ratios have been on the rise and now above the norm. I have found through the years my best returns have come from buying during downdrafts when stocks are, at times, selling back of normal p/e ratios.
Reply to @MarkM: Interesting. Did you really see a tautology or is it circular reasoning? Care to explain? It should be fairly easy to point out the logical fallacy. Wouldn't be surprised if that is where the predictive capabilities come from.
Anyway, I am not sure he is saying markets are logical. He is trying to make a logical argument that the markets are behavioral and valuations based on money chasing assets, not an earth shattering thesis. Of course, one might say that PPT acts by influencing the supply of money amongst various conspiracy theories in which case it has been considered already.
Reply to @MarkM: I must admit this link was long winded...MJG might approve, but I had to chew...swallow...chew...swallow...and somewhere along the way I stopped chewing.
Reply to @cman: Also the the velocity of money has been pretty muted...banks with higher lending standards and individuals not willing to take on too much additional debt. Even today's car loans seem far below historical rates.
Are you suggesting there is a basic flaw in his presentation that fully eliminates or diminishes the potential predictive usefulness of the metric he presents? If so, it would be appreciated if you would point out the flaw.
This metric is quite interesting and makes intuitive sense to me. The 10 year time horizon is long enough for accepted market norms to change -- for “market memory” to be reset at a fairly deep level. [Its the cutoff point William Bernstein picked as the starting point in his transition from shallow investment risk to deep investment risk – where deep investment risk is viewed as a “permanent” loss of capital (Deep Risk: How History Informs Portfolio Design).]
The scatter plot suggests the metric has done a good job of “predicting” 10 year average returns over the past half century. It also suggests we may currently be somewhere in the middle area for potential returns for the next decade....not at one extreme or the other.
David's January commentary discussed relative value and absolute value stock investors. I tend towards the absolute value camp and have about 50% of my investment portfolio invested with stock managers who adhere to this philosophy (with FPACX included). The metric presented with this post suggests it may be a few years before adherents to an absolute value style of investing have their next full day in the sun...relatively speaking!
Let me start with a disclaimer. I breezed through the referenced article since most of it is “old” stuff and commonly accepted market wisdom. So I do not claim that my posting is a definitive or complete review of the paper.
Of course, the idea to replace the conventional portion of what the author elected to call the “Price Return” with investor equity percentage growth commitment and cash-bond supply growth components is dramatic and inventive. I like that. Only time will test the validity and usefulness of the proposed concept. But many kudos for trying. I’ll mostly focus my comments on this aspect of the paper.
Take note that even the basic partitioning of the initial total equity returns equation is a departure from the standard sectioning. In John Bogle’s classic “Common Sense on Mutual Funds” book, he divides the total returns into investment and speculative subcategories. Bogle assigns earnings growth rate and dividend yield to the investment category, and reserves P/E ratio changes to the speculative category. This researcher doesn’t accept that interpretation.
I suppose it somewhat depends on your investment philosophy. Bogle is certainly conservative and long-term oriented. The current writer’s proclivities are unknown since he chooses to be anonymous.
This fact itself is a little troublesome. If this thesis were generated at the University of Chicago or MIT, it would be critically peer reviewed. Not so in this instance. If in time this provisional model proves to be a flawed concept, the unknown author will simply fade away and not pay any reputation price for his faulty analyses.
On the positive side, the proposed model does take advantage of two commonly acknowledged heuristics: the wisdom of the crowds and a regression-to-the-mean. Both are used in an inverted sense. The wisdom of the crowd is reflected as the independent variable in the proposed model. It is the worldwide sum of all equity investors. However, they are typically on the wrong side of the wager. When they are heavily invested, the equity marketplace is likely overbought and a reversion-to-the-mean is the order of the day.
The reported work sets a low hurdle for itself. It uses the current value of investor enthusiasm, as measured by their equity percentage holdings, to project a 10-year rolling return. The author sort of endows the investor with a prescience that is hard to justify. Why not a 1 year forecast, or a 3-year, or a 5-year, or a 20-year timeframe? Data mining is a possibility in this instance.
Another consideration is that a meaningful test of the validity of the asserted model must wait and wait and wait and wait before seeing if the model bears fruit. Immediate feedback is impossible. Therefore, the learning process is delayed.
I was disappointed that the author elected not to challenge his model with out-of-sample data. For example, he could have used data sets before World War II. High quality data of the form needed might not have been available, so, as an alternative approach, this researcher could have segregated his post WW II data into subsets and tested the accuracy of the resultant forecasts for each subset.
Out-of-sample testing is always needed. Researchers always reserve data for that purpose. We’re all familiar with the failures of the long skirt, the super-bowl, and the Bangladesh butter production correlations to forecast equity returns when challenged with evolving data. More seriously formulated correlations like the Shiller CAPE model worked for a while and then failed. The investment roadway is littered with such failed correlations. Caution is warranted.
Forecasters fail to forecast; there is little reason to believe that equity investors possess that skill. Black swans and exogenous events happen to destroy any hopes for accurate long range forecasts. The author takes some pride that the proposed forecasting tool does not incorporate fundamental market parameters. I tend to distrust such a correlation as a weakness.
The author elects not to emphasize the data scatter about his correlation until the very end of the article. That scatter (standard deviation) is always an important consideration in any investment decision. Currently, based on the proposed correlation, the equity markets offer a 10-year rolling average return of 6 % with data showing scatter between a 5 % and 9 % level. That’s not a particularly shocking projection.
As the forecast time horizon expands, the projections should reflect the historical averages. Essentially, that’s what I meant when I said that the author set himself a low hurdle. As a zeroth order prediction I could simply quote the historical averages and its standard deviation. As a first order correction to that initial estimate, I could correct either upward or downward using CAPE as a guide. This procedure is just a reversion-to-the-mean generic philosophy and might well be as reliable as the referenced research.
Only time will tell. Unfortunately, given the time horizon for the referenced work, that wait will be painfully long. I have no horses in this race. I really do wish the author well. I like attempts to improve our market understanding. Most models fail; especially since investor sentiment is fragile and ever changing, it is a tough modeling nut.
I hope you guys find this a little helpful. Please remember that I have not done a meticulous scrutiny of the referenced article. Daniel Kahneman would not be a happy warrior with my submittal given that it is almost solely based on my reflexive thinking on the matter.
Comments
FRED
His premise is similar to my thesis posted earlier in the context of discussing stock multiples. That stocks have become derivatives in the direction based on future investor availability than any fundamentals while we use the latter to keep a pretense that it is not.
I have not gone through his quantitative model yet to see if it makes sense and skeptical that there is predictive capability to this thesis but I am keeping an open mind.
Thanks for bringing this to the board. At first read I follow a good bit of the thinking and logic that it expresses. However, I am not so sure the markets follow logic and there is a group called the Plunge Protection Team (PPT). The PPT can have a great effect on the markets as policy changes within the FOMC itself. With this, these policies and changes thereof are directed to the Federal Reserve and its member banks. Some of the key figures that sit on the PPT come from some of the country's largest banks and also include the Treasury, the FOMC chairperson and others.
For those interested I have linked more information below on the PPT.
http://en.wikipedia.org/wiki/Working_Group_on_Financial_Markets
Now, how does one factor in all this?
Perhaps, our markets are more influenced by the banks than ones wishes to believe. In any event if the markets are trading above normal p/e ratios then overall returns will be less by my thinking. This is one of the reasons I have been reducing my allocation to equities over the past couple of years as p/e ratios have been on the rise and now above the norm. I have found through the years my best returns have come from buying during downdrafts when stocks are, at times, selling back of normal p/e ratios.
Old_Skeet
Anyway, I am not sure he is saying markets are logical. He is trying to make a logical argument that the markets are behavioral and valuations based on money chasing assets, not an earth shattering thesis. Of course, one might say that PPT acts by influencing the supply of money amongst various conspiracy theories in which case it has been considered already.
Are you suggesting there is a basic flaw in his presentation that fully eliminates or diminishes the potential predictive usefulness of the metric he presents? If so, it would be appreciated if you would point out the flaw.
This metric is quite interesting and makes intuitive sense to me. The 10 year time horizon is long enough for accepted market norms to change -- for “market memory” to be reset at a fairly deep level. [Its the cutoff point William Bernstein picked as the starting point in his transition from shallow investment risk to deep investment risk – where deep investment risk is viewed as a “permanent” loss of capital (Deep Risk: How History Informs Portfolio Design).]
The scatter plot suggests the metric has done a good job of “predicting” 10 year average returns over the past half century. It also suggests we may currently be somewhere in the middle area for potential returns for the next decade....not at one extreme or the other.
David's January commentary discussed relative value and absolute value stock investors. I tend towards the absolute value camp and have about 50% of my investment portfolio invested with stock managers who adhere to this philosophy (with FPACX included). The metric presented with this post suggests it may be a few years before adherents to an absolute value style of investing have their next full day in the sun...relatively speaking!
I have the suspicion that A-A reasoning is being used as well. It would take Dr. Hussman to refute.
However, it logically makes sense. Once more investors allocate to stock the P/E ratio goes up and stock become overvalued.
Let me start with a disclaimer. I breezed through the referenced article since most of it is “old” stuff and commonly accepted market wisdom. So I do not claim that my posting is a definitive or complete review of the paper.
Of course, the idea to replace the conventional portion of what the author elected to call the “Price Return” with investor equity percentage growth commitment and cash-bond supply growth components is dramatic and inventive. I like that. Only time will test the validity and usefulness of the proposed concept. But many kudos for trying. I’ll mostly focus my comments on this aspect of the paper.
Take note that even the basic partitioning of the initial total equity returns equation is a departure from the standard sectioning. In John Bogle’s classic “Common Sense on Mutual Funds” book, he divides the total returns into investment and speculative subcategories. Bogle assigns earnings growth rate and dividend yield to the investment category, and reserves P/E ratio changes to the speculative category. This researcher doesn’t accept that interpretation.
I suppose it somewhat depends on your investment philosophy. Bogle is certainly conservative and long-term oriented. The current writer’s proclivities are unknown since he chooses to be anonymous.
This fact itself is a little troublesome. If this thesis were generated at the University of Chicago or MIT, it would be critically peer reviewed. Not so in this instance. If in time this provisional model proves to be a flawed concept, the unknown author will simply fade away and not pay any reputation price for his faulty analyses.
On the positive side, the proposed model does take advantage of two commonly acknowledged heuristics: the wisdom of the crowds and a regression-to-the-mean. Both are used in an inverted sense. The wisdom of the crowd is reflected as the independent variable in the proposed model. It is the worldwide sum of all equity investors. However, they are typically on the wrong side of the wager. When they are heavily invested, the equity marketplace is likely overbought and a reversion-to-the-mean is the order of the day.
The reported work sets a low hurdle for itself. It uses the current value of investor enthusiasm, as measured by their equity percentage holdings, to project a 10-year rolling return. The author sort of endows the investor with a prescience that is hard to justify. Why not a 1 year forecast, or a 3-year, or a 5-year, or a 20-year timeframe? Data mining is a possibility in this instance.
Another consideration is that a meaningful test of the validity of the asserted model must wait and wait and wait and wait before seeing if the model bears fruit. Immediate feedback is impossible. Therefore, the learning process is delayed.
I was disappointed that the author elected not to challenge his model with out-of-sample data. For example, he could have used data sets before World War II. High quality data of the form needed might not have been available, so, as an alternative approach, this researcher could have segregated his post WW II data into subsets and tested the accuracy of the resultant forecasts for each subset.
Out-of-sample testing is always needed. Researchers always reserve data for that purpose. We’re all familiar with the failures of the long skirt, the super-bowl, and the Bangladesh butter production correlations to forecast equity returns when challenged with evolving data. More seriously formulated correlations like the Shiller CAPE model worked for a while and then failed. The investment roadway is littered with such failed correlations. Caution is warranted.
Forecasters fail to forecast; there is little reason to believe that equity investors possess that skill. Black swans and exogenous events happen to destroy any hopes for accurate long range forecasts. The author takes some pride that the proposed forecasting tool does not incorporate fundamental market parameters. I tend to distrust such a correlation as a weakness.
The author elects not to emphasize the data scatter about his correlation until the very end of the article. That scatter (standard deviation) is always an important consideration in any investment decision. Currently, based on the proposed correlation, the equity markets offer a 10-year rolling average return of 6 % with data showing scatter between a 5 % and 9 % level. That’s not a particularly shocking projection.
As the forecast time horizon expands, the projections should reflect the historical averages. Essentially, that’s what I meant when I said that the author set himself a low hurdle. As a zeroth order prediction I could simply quote the historical averages and its standard deviation. As a first order correction to that initial estimate, I could correct either upward or downward using CAPE as a guide. This procedure is just a reversion-to-the-mean generic philosophy and might well be as reliable as the referenced research.
Only time will tell. Unfortunately, given the time horizon for the referenced work, that wait will be painfully long. I have no horses in this race. I really do wish the author well. I like attempts to improve our market understanding. Most models fail; especially since investor sentiment is fragile and ever changing, it is a tough modeling nut.
I hope you guys find this a little helpful. Please remember that I have not done a meticulous scrutiny of the referenced article. Daniel Kahneman would not be a happy warrior with my submittal given that it is almost solely based on my reflexive thinking on the matter.
Best Regards.