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.
I have to agree with Ed Studzinski that at present "cash is the undervalued asset."
I've felt that way for some time now. Glad I didn't follow my own advice, as I'm sitting on a double-digit return year to date. Cash wouldn't have done it. Still, I see the logic in Ed's comment.
Some real good stuff in David's monthly publication. Read it if you haven't already.
Agreed. I haven't "raised cash" this year and am letting my various holdings ride as they will but still have a sizable amount that I've yet to find good values to put it to work on. Ergo, it remains liquid, and I remain patient ... which presumably makes me a bad investor in the eyes of Wall Street.
Cash is a sizeable amount within my portfolio's asset allocation with a range of 15% to 25%. According to a recent Xray analysis it is at its upper limit at 25% without being overweight cash. In the nearterm, it might just become overweight.
Like Ed's comment too on cash, but I see it as opportunities to repositioning the portfolio at the right time. Little did I know that emerging market is the leading asset class this year and Andrew Foster came through again.
Robert Cochran's column, in its use of mean reversion and inflation/real return, has left me befuddled.
"Reversion to the mean" simply expresses the tendency of next year's returns to be closer to the long term mean than the current year's returns are.
Underlying mean reversion is the assumption that each year's performance is independent of the previous one's. That is, mean reversion applies to random variables.
Assuming a long term mean of 9-10% for stocks (as stated in the opening sentence), and assuming 2016's return comes out about 12% (extrapolating from 8% YTD), mean reversion suggests that it is more likely next year's returns will be lower (closer to the mean of 10%) than higher (further from the mean). That's all.
Many prognosticators suggest that stock returns going forward will average around 4-5% (with an assortment of solid reasons backing this up). Mean reversion would seem to cut against this, as it implies, quite literally, reversion (coming closer) to the mean of 10%. IMHO this just shows that mean reversion doesn't apply here - yearly returns are not random variables.
Regarding real returns and inflation - if inflation is assumed to run at 2-3% (it isn't now, but it is expected to increase), then SS should also increase in nominal terms 2-3%, not the 1% projected. In real terms (as measured by CPI-W), SS payments do not decrease.
The 5% average figure for bonds over the past 15 years suggests that "bonds" means 10 year bonds. See here (geometric average over past ten years was 4.71% for 10 year bonds). That same source also shows an average near 5% (4.96%) for the past 85 years. Arithmetic averages are similar, though slightly higher (a small fraction above 5%).
So it seems fair to use last century's (100 year) average real returns for 10 year bonds as "normal" returns. That average real return was around 1.6% in the US:
If you prefer, 1.7% real return for 1900-2002 (based on Shiller data)
So I don't understand what the big deal is about a 0-2% real return going forward. That sounds about normal.
If anything, achieving typical real returns with lower nominal returns and lower inflation is beneficial to fixed income investors. That's because taxes are based on nominal returns, not real returns. So achieving the same real returns and paying less in taxes (lower nominal returns) seems like a plus.
In the broad picture, I agree with the expectation that both stock and bond returns will be lower going forward. But not as explained. Stocks may violate mean reversion (i.e. overshoot the mean on the low side, rather than simply dropping closer to the mean). Parallel increases in prices and rates would keep bond real returns closer to zero.
Rhetorical tropes are tricky. I've read apophasis as more allusive than paralipsis. "I won't mention your misadventures as a young adult" as opposed to "we needn't dwell on my opponent's criminal past, neither will we speak of her convenient lapses of memory nor on her repeated uses of the word 'nu cu lar' ...". Others claim that the distinction is too fine (though when that would have deterred a classical rhetorician is unclear to me) and the words should be treated as synonyms.
I used to teach rhetorical theory but I was at pains to avoid inflicting lists of figures and tropes of the kids. We might debate whether rhetoric was a way of discovering a truth or of presenting it, but I was always sensitive to Samuel Butler's gibe: "For all a rhetorician's rules. Teach nothing but to name his tools."
Hi, msf. Given the fact that we have had out-sized returns for the S&P 500 the last 5 years (average of about 15.5%), with some sectors much, much higher, it is natural to expect that we could well have some lean years if longer-term average numbers are to be trusted. The 10-year S&P 500 average return is only 7.4%, a long way from the outrageously long historical number, which some retirement web sites still allow using. So if we are to have future average returns of around 7%, there will need to be some very poor years to bring the market average down to that level. Or we could have one or two awful years. Perhaps the need to keep words to a minimum meant a deeper or clearer explanation was left out. I hope this clears the water.
As for bond yields, I think the fact that we are in totally uncharted waters with interest rates might result in strongly negative returns for bonds. I am not aware that so many countries have ever suckered poor souls to buy bonds with negative yields. And while U.S. yields are higher than 0%, many bond prices are so high as to suggest owners could have negative returns if rates move up by just 1%.
I am not suggesting returns for stocks or bonds is about to be hideous, but I do believe that using an assumed average return of more than 4-5% for retirement projections is unwise.
Hi Bob, thanks for the comments. I completely agree with you that after several fat cows one should expect several lean ones. But mean regression is something entirely different.
What you (and I, and most everyone else) are assuming is that there's some relationship between past and future (lean follows fat). Mean regression assumes the complete opposite - that each year is completely independent and random.
Now you don't believe that next year is disconnected from this year or the past few. Neither do I. So mean regression is not applicable. Even if it were, it never predicts bad years, just that if this year was good, next year will, more likely than not, be less good.
With respect to bonds ... If one buys a bond now, even a premium bond, whether the real rate of return turns out to be positive or not depends on rate of inflation until maturity. It doesn't matter whether nominal rates go up 1% next year, that's not going to affect the coupons, the return of principal, or the real return. (Except arguably by inference that inflation may rise in tandem with the rise in nominal rates - see Fisher hypothesis.)
Inflation is still hovering below 1%. (0.8% Y/Y as of July - see graph here). Target rate is 2%. If we approach that without overshooting (admittedly a significant assumption), then 10 year bonds, yielding 1.60% nominal as of 9/2/16 should generate a small but positive real return. It doesn't matter what happens to market rates; what matters once the investment is made is the rate of inflation.
Still, there's a difference between this barely positive real return and the historical real return of 1.6%. So I agree that one needs to plan for lower returns than the historical average for both bonds and stocks. Though it remains important to be clear on the reasons for that qualitative projection, in order to make good quantitative guesses.
I think your 4-5% (nominal) is a good, conservative figure for planning purposes. IMHO that puts real return somewhere around 2%.
In the broad picture, I agree with the expectation that both stock and bond returns will be lower going forward. But not as explained. Stocks may violate mean reversion (i.e. overshoot the mean on the low side, rather than simply dropping closer to the mean). Parallel increases in prices and rates would keep bond real returns closer to zero.
I agree with this. The reversion to the mean is better applied to probabilities (e.g. coin toss, roulette ) and requires large numbers. None of those things apply with the stock market.
We've had lower than normal GDP growth recently and that will likely continue as was discussed in the thread on productivity recently.
@msf >> Mean regression assumes ... that each year is completely independent and random. >> ... mean regression predicts that if this year was good, next year will, more likely than not, be less good.
? - meaning there is no inertia to this year's goodness, and therefore independence necessarily leads to the expectation that next year will probably be not the same as to goodness.
Is there a technically acceptable term for what most people mean by reversion to a norm? Ditto for persistence of sameness?
Your desire to learn more on what quickly becomes a complex mathematically dominated topic is highly commendable and deserves respect. Good luck in your exploration. I abandoned my interest a few years ago because of my own persistency shortfalls.
The subject has been studied for decades from an investment enhancing perspective. It fundamentally explores the persistency of momentum in the marketplace for various timeframes: intraday, daily, monthly, quarterly, and on an annual basis.
The research seemed to demonstrate that on any time scale markets were never completely random; some momentum, some local dependence existed for rapidly disappearing time periods. A perfectly independent event would have a zero correlation. The numerous studies always showed values slightly departing from that zero. These numbers never seemed to depart by very much, but the researchers tested them and concluded they were not random noise.
These studies are called autocorrelations. They are linear regression analyses with some period time lag used in the self-comparison. The findings vary over various timeframes. They were dynamic and not constant. That's a warning signal.
I've not been impressed that these results generate actionable investment opportunities. If they did, we would all know about them.
I hope this helps. Note that I am not an expert in this field.
As a technical editor I was just interested in getting my terms and understanding right, or less wrong, investing aside. Independent events appear to correlate all the time, so it's the parsing of that over time that I wish to understand. wikip has good explans of linear (and nonlinear) regression, which I have had to consult editing casually spoken engineers' Darpa proposals.
While MJG was posting his response, I'd drafted this. At least I didn't seem to get anything wrong:
Often one models stock returns as periodic time series using autoregression - that is, treating annual returns as a pattern that approximately repeats. Simplistically, the next return is essentially a "copy" of the stock's return a full cycle ago. Of course cycles are rarely that simple, and there's always noise. Or as Mark Twain put it, history doesn't repeat, but it does rhyme.
But that's the model. What one calls the tendency to track a cycle (e.g. move from above average returns to below average returns), I don't know. Statistics is a subject I avoided in school, though work has pulled me quite a bit in its direction.
Likewise, nothing much comes to mind for persistence of sameness. Inertia? "New normal?"
Sorry - hope someone better schooled can come up with the precise terms.
I hate hitting on a small point, but I post now with a minor complaint. You do DARPA an injustice by not using Caps uniformly.
DARPA is the Defense Advanced Research Project Agency. It was formed in 1958 in response to an emerging Soviet space threat. It's superior performance record is unmatched by any other government institution or by industry from which it draws members to establish teams with limited assignments. This agency developed the Internet.
Good for them and great for us. They deserve a proper recognition and acknowledgement for their long and successful service record. Three cheers for DARPA.
The longer proper-name acronyms (meaning pronounceable) have a long and honorable history of going initial-cap only (optionally). Cf. Nasdaq, Unicef, Nascar among many others. This minor language issue aside, you never need wag your finger about such at someone who has worked for over a decade doing Darpa proposals. (And as a former B B N staffer I can tell you for sure that Darpa did not develop the Internet.)
We ought to stop doing this or folks will start talking.
Sorry that we disagree on who should get the most credit for the invention of such a useful worldwide product as the Internet. The concept did develop over an extended timeframe, likely starting as early as with Tesla. Lots of warranted ownership claims, but also lots of false claims.
One expert said that " the first workable prototype of the Internet came in the late 1960s with the creation of ARPANET." The ARPA name was later changed to DARPA to reflect the changing world dynamic.
I too have had the good fortune to work on DARPA teams on several projects, but that was several decades ago. DARPA carefully selects their team members wisely. I really did feel that I contributed to some sound assessments and decisions.
I stand firm. It is DARPA and not Darpa. That's my opinion and that's what MFO is all about. You are always free to disagree and to express that disagreement. MFO members get to support whoever presents a more compelling case.
EDIT: Here is the Link that I extracted the quote from:
I made my original claim of discovery from memory. The DARPA guys told me it was so. I acknowledge that the debate has not and will likely never end. Lots of folks have added major additions to the system's functionality.
One final comment. On its own website, DARPA presents itself in full Caps. Here is a Link to that site:
"Your desire to learn more on what quickly becomes a complex mathematically dominated topic is highly commendable and deserves respect." That's what all my teachers used to say (or something like that)
Great discussion. I guess what investors really need to look at is the real return on different competing types of investments over long periods. How have various assets, including equities, fared compared to the cost of living over time? (Of course, cost of living itself is subject to different methods of quantification.)
Example: A nominal 5% annual return over the past decade probably would have kept most of us ahead or about even with COL. But in the 70s, a 5% annual return would have resulted in rapidly deteriorating living standards for most.
Still, the task would seem daunting, since different assets are likely to perform very differently over extended periods of time and in different environments. Example: Return on treasury bonds over the past decade (and probably longer) would have likely kept most ahead of COL. But in the 70s the reverse would have been true.
Not a debate, not in anyone's mind. Everyone acknowledges by now the team, the players, the protocols, the state of art by stages, and the historical sequence. You said develop. Development in the usual senses was not then and is not now Darpa's role. Your quote illustrates perfectly, the opening stages. Read the wikip article for more. For some reason you are acting as though I am slighting Darpa by going with common newspaper usage and also by trying to explicate for you the civilian role in development of mil initiatives and demonstration prototypes.
My bosses (and some colleagues) at both B B N and QinetiQ / Foster-Miller were mostly ex-Darpa PMs or headed that way soon after.
Seems like the to time to be "steering" investors towards indexing, passive, and ultra diversification is when markets have been through a reasonable decline. Not at a time when: 1) the largest stock market ( U.S.) in the world is richly valued and forward return expectations as measured by many metrics are low 2) many world bond yields are sparse 3) the average investor within a few years of retirement age, who are deficient in retirement asset accumulation ( a large percentage ), will need some sort of high alpha, active / strategic & capital preservation portfolios in order to "catch up" and maintain a reasonable retirement lifestyle.
Don't hear anyone pushing for an overweight in emerging market / European bourses either which would seem to be logical and inverse to point #1.
Thank you for the excellent reference. It was more than I ever cared to know about DARPA.'s early history.
In fact, it reinforced my basic comments.
Look, it's a small, no a tiny matter. Throughout your referenced Link, ARPA specified itself as ARPA and not Arpa. To paraphrase one of Norm Augustine's many laws, this exchange started slowly, and then sort of tapered off. I suspect neither you or me will alter our positions. Again, it's a very tiny matter, but I interpret it as disrespectful to DARPA.
Comments
I've felt that way for some time now. Glad I didn't follow my own advice, as I'm sitting on a double-digit return year to date. Cash wouldn't have done it. Still, I see the logic in Ed's comment.
Some real good stuff in David's monthly publication. Read it if you haven't already.
Agreed. I haven't "raised cash" this year and am letting my various holdings ride as they will but still have a sizable amount that I've yet to find good values to put it to work on. Ergo, it remains liquid, and I remain patient ... which presumably makes me a bad investor in the eyes of Wall Street.
"Reversion to the mean" simply expresses the tendency of next year's returns to be closer to the long term mean than the current year's returns are.
Underlying mean reversion is the assumption that each year's performance is independent of the previous one's. That is, mean reversion applies to random variables.
Assuming a long term mean of 9-10% for stocks (as stated in the opening sentence), and assuming 2016's return comes out about 12% (extrapolating from 8% YTD), mean reversion suggests that it is more likely next year's returns will be lower (closer to the mean of 10%) than higher (further from the mean). That's all.
Many prognosticators suggest that stock returns going forward will average around 4-5% (with an assortment of solid reasons backing this up). Mean reversion would seem to cut against this, as it implies, quite literally, reversion (coming closer) to the mean of 10%. IMHO this just shows that mean reversion doesn't apply here - yearly returns are not random variables.
Regarding real returns and inflation - if inflation is assumed to run at 2-3% (it isn't now, but it is expected to increase), then SS should also increase in nominal terms 2-3%, not the 1% projected. In real terms (as measured by CPI-W), SS payments do not decrease.
The 5% average figure for bonds over the past 15 years suggests that "bonds" means 10 year bonds. See here (geometric average over past ten years was 4.71% for 10 year bonds). That same source also shows an average near 5% (4.96%) for the past 85 years. Arithmetic averages are similar, though slightly higher (a small fraction above 5%).
So it seems fair to use last century's (100 year) average real returns for 10 year bonds as "normal" returns. That average real return was around 1.6% in the US:
If you prefer, 1.7% real return for 1900-2002 (based on Shiller data)
So I don't understand what the big deal is about a 0-2% real return going forward. That sounds about normal.
If anything, achieving typical real returns with lower nominal returns and lower inflation is beneficial to fixed income investors. That's because taxes are based on nominal returns, not real returns. So achieving the same real returns and paying less in taxes (lower nominal returns) seems like a plus.
In the broad picture, I agree with the expectation that both stock and bond returns will be lower going forward. But not as explained. Stocks may violate mean reversion (i.e. overshoot the mean on the low side, rather than simply dropping closer to the mean). Parallel increases in prices and rates would keep bond real returns closer to zero.
I used to teach rhetorical theory but I was at pains to avoid inflicting lists of figures and tropes of the kids. We might debate whether rhetoric was a way of discovering a truth or of presenting it, but I was always sensitive to Samuel Butler's gibe: "For all a rhetorician's rules. Teach nothing but to name his tools."
David
As for bond yields, I think the fact that we are in totally uncharted waters with interest rates might result in strongly negative returns for bonds. I am not aware that so many countries have ever suckered poor souls to buy bonds with negative yields. And while U.S. yields are higher than 0%, many bond prices are so high as to suggest owners could have negative returns if rates move up by just 1%.
I am not suggesting returns for stocks or bonds is about to be hideous, but I do believe that using an assumed average return of more than 4-5% for retirement projections is unwise.
What you (and I, and most everyone else) are assuming is that there's some relationship between past and future (lean follows fat). Mean regression assumes the complete opposite - that each year is completely independent and random.
Now you don't believe that next year is disconnected from this year or the past few. Neither do I. So mean regression is not applicable. Even if it were, it never predicts bad years, just that if this year was good, next year will, more likely than not, be less good.
With respect to bonds ... If one buys a bond now, even a premium bond, whether the real rate of return turns out to be positive or not depends on rate of inflation until maturity. It doesn't matter whether nominal rates go up 1% next year, that's not going to affect the coupons, the return of principal, or the real return. (Except arguably by inference that inflation may rise in tandem with the rise in nominal rates - see Fisher hypothesis.)
Inflation is still hovering below 1%. (0.8% Y/Y as of July - see graph here). Target rate is 2%. If we approach that without overshooting (admittedly a significant assumption), then 10 year bonds, yielding 1.60% nominal as of 9/2/16 should generate a small but positive real return. It doesn't matter what happens to market rates; what matters once the investment is made is the rate of inflation.
Still, there's a difference between this barely positive real return and the historical real return of 1.6%. So I agree that one needs to plan for lower returns than the historical average for both bonds and stocks. Though it remains important to be clear on the reasons for that qualitative projection, in order to make good quantitative guesses.
I think your 4-5% (nominal) is a good, conservative figure for planning purposes. IMHO that puts real return somewhere around 2%.
We've had lower than normal GDP growth recently and that will likely continue as was discussed in the thread on productivity recently.
>> Mean regression assumes ... that each year is completely independent and random.
>> ... mean regression predicts that if this year was good, next year will, more likely than not, be less good.
? - meaning there is no inertia to this year's goodness, and therefore independence necessarily leads to the expectation that next year will probably be not the same as to goodness.
Is there a technically acceptable term for what most people mean by reversion to a norm? Ditto for persistence of sameness?
Your desire to learn more on what quickly becomes a complex mathematically dominated topic is highly commendable and deserves respect. Good luck in your exploration. I abandoned my interest a few years ago because of my own persistency shortfalls.
The subject has been studied for decades from an investment enhancing perspective. It fundamentally explores the persistency of momentum in the marketplace for various timeframes: intraday, daily, monthly, quarterly, and on an annual basis.
The research seemed to demonstrate that on any time scale markets were never completely random; some momentum, some local dependence existed for rapidly disappearing time periods. A perfectly independent event would have a zero correlation. The numerous studies always showed values slightly departing from that zero. These numbers never seemed to depart by very much, but the researchers tested them and concluded they were not random noise.
These studies are called autocorrelations. They are linear regression analyses with some period time lag used in the self-comparison. The findings vary over various timeframes. They were dynamic and not constant. That's a warning signal.
I've not been impressed that these results generate actionable investment opportunities. If they did, we would all know about them.
I hope this helps. Note that I am not an expert in this field.
Best Wishes.
Often one models stock returns as periodic time series using autoregression - that is, treating annual returns as a pattern that approximately repeats. Simplistically, the next return is essentially a "copy" of the stock's return a full cycle ago. Of course cycles are rarely that simple, and there's always noise. Or as Mark Twain put it, history doesn't repeat, but it does rhyme.
But that's the model. What one calls the tendency to track a cycle (e.g. move from above average returns to below average returns), I don't know. Statistics is a subject I avoided in school, though work has pulled me quite a bit in its direction.
Likewise, nothing much comes to mind for persistence of sameness. Inertia? "New normal?"
Sorry - hope someone better schooled can come up with the precise terms.
I hate hitting on a small point, but I post now with a minor complaint. You do DARPA an injustice by not using Caps uniformly.
DARPA is the Defense Advanced Research Project Agency. It was formed in 1958 in response to an emerging Soviet space threat. It's superior performance record is unmatched by any other government institution or by industry from which it draws members to establish teams with limited assignments. This agency developed the Internet.
Good for them and great for us. They deserve a proper recognition and acknowledgement for their long and successful service record. Three cheers for DARPA.
Best Wishes.
We ought to stop doing this or folks will start talking.
Sorry that we disagree on who should get the most credit for the invention of such a useful worldwide product as the Internet. The concept did develop over an extended timeframe, likely starting as early as with Tesla. Lots of warranted ownership claims, but also lots of false claims.
One expert said that " the first workable prototype of the Internet came in the late 1960s with the creation of ARPANET." The ARPA name was later changed to DARPA to reflect the changing world dynamic.
I too have had the good fortune to work on DARPA teams on several projects, but that was several decades ago. DARPA carefully selects their team members wisely. I really did feel that I contributed to some sound assessments and decisions.
I stand firm. It is DARPA and not Darpa. That's my opinion and that's what MFO is all about. You are always free to disagree and to express that disagreement. MFO members get to support whoever presents a more compelling case.
EDIT: Here is the Link that I extracted the quote from:
http://www.history.com/news/ask-history/who-invented-the-internet
I made my original claim of discovery from memory. The DARPA guys told me it was so. I acknowledge that the debate has not and will likely never end. Lots of folks have added major additions to the system's functionality.
One final comment. On its own website, DARPA presents itself in full Caps. Here is a Link to that site:
http://www.darpa.mil/
That's enough said.
Best Wishes.
Great discussion. I guess what investors really need to look at is the real return on different competing types of investments over long periods. How have various assets, including equities, fared compared to the cost of living over time? (Of course, cost of living itself is subject to different methods of quantification.)
Example: A nominal 5% annual return over the past decade probably would have kept most of us ahead or about even with COL. But in the 70s, a 5% annual return would have resulted in rapidly deteriorating living standards for most.
Still, the task would seem daunting, since different assets are likely to perform very differently over extended periods of time and in different environments. Example: Return on treasury bonds over the past decade (and probably longer) would have likely kept most ahead of COL. But in the 70s the reverse would have been true.
2-cents worth
Jeez, so easy to check what actually happened and how it all went:
https://en.wikipedia.org/wiki/ARPANET
Not a debate, not in anyone's mind. Everyone acknowledges by now the team, the players, the protocols, the state of art by stages, and the historical sequence. You said develop. Development in the usual senses was not then and is not now Darpa's role. Your quote illustrates perfectly, the opening stages. Read the wikip article for more.
For some reason you are acting as though I am slighting Darpa by going with common newspaper usage and also by trying to explicate for you the civilian role in development of mil initiatives and demonstration prototypes.
My bosses (and some colleagues) at both B B N and QinetiQ / Foster-Miller were mostly ex-Darpa PMs or headed that way soon after.
1) the largest stock market ( U.S.) in the world is richly valued and forward return expectations as measured by many metrics are low
2) many world bond yields are sparse
3) the average investor within a few years of retirement age, who are deficient in retirement asset accumulation ( a large percentage ), will need some sort of high alpha, active / strategic & capital preservation portfolios in order to "catch up" and maintain a reasonable retirement lifestyle.
Don't hear anyone pushing for an overweight in emerging market / European bourses either which would seem to be logical and inverse to point #1.
Thank you for the excellent reference. It was more than I ever cared to know about DARPA.'s early history.
In fact, it reinforced my basic comments.
Look, it's a small, no a tiny matter. Throughout your referenced Link, ARPA specified itself as ARPA and not Arpa. To paraphrase one of Norm Augustine's many laws, this exchange started slowly, and then sort of tapered off. I suspect neither you or me will alter our positions. Again, it's a very tiny matter, but I interpret it as disrespectful to DARPA.
Stay cool and stay strong.
Best. Wishes.