Bogleheads

[larryswedroe]

Henry



Risks are
a) tracking error
b) historical results for beta, size and value premiums will look different than in past with beta being larger and the others being smaller (of course reverse might be true)

[Sammy_M]

[Jimmy Snuka]

[Henry Sadovsky]

[larryswedroe]

Henry
Also less "risk" of achieving a sizable portfolio--of course trade off is less risk of ending up with very small portfolio

And as I said the risk is also that beta can turn out to be high and small and value way underperform, so you don't get the expected return--but should not care because that confuses strategy and outcome--the idea is to reduce the risk of the left tail.

[Roy]

[Henry Sadovsky]

.

[larryswedroe]

Roy an awful lot, very high percentage.

When I ran the data a while ago about 44% equity exposure for a model portfolio we use at our firm had same returns and about 1/3 the SD of TSM. And my portfolio is a lot more tilted.

Henry--don't agree. Difference is you posit risk as volatility. Volatility is only one type of risk. If it was only definition then CAPM would have remained the model (one factor) but we have three factor which does much better job of explaining returns.

One of the benefits is that the three factors are unique or independent risks, with very low correlations, which serves to lower the volatility of the portfolio.

BTW-there are many papers that show the logical risk explanations of the value premium (covered in my books) and no one questions the size premium as a risk story. And if the value premium is even part behavioral then there is a free lunch there. So what exactly do you think I am hoping for? (that there really is a value premium? that the low correlations will go way up?) And if I am wrong what have I really lost? Remember my objective is to reduce left tail risk and am willing to give up right tail possibility.

Personally I dont think this is much of a gamble at all. IMO there are logical risk stories here and there are unique risk factors (low correlation) and it seems highly unlikely that my portfolio could possibly have negative returns while a high equity portfolio would have large gains, and also I find it hard to believe that my portfolio can suffer drastic losses without beta being big negative--so I win in that world too in almost certainty. Remember the objectives here are different.

best

[Henry Sadovsky]

.

[james22]

[larryswedroe]

Henry
First,
I dont have time to write about all the papers that provide very strong and logical explanations for the value premiums that are economic based. You can find them in my books if interested.

Second, don't know how else to say this but risks doesn't have to be volatility. That is simply one definition of risk. As I said if it was the only one we would have a one factor world

Third, I don't know why you keep repeating the same statement when I have already said very explicitly that one risk is that my estimates of the risk factor returns turns out differently than expected. If that were not the case then they would not be risk factors would they? They would be certainties. I accept that risk. But as I said, my objective is to avoid the left tail. You show me a likely outcome where I have big losses but a much higher equity allocation does not. Never happened and I cannot see how it is likely. Now the right tail can easily happen and has happened but I care little about that as I am trying to avoid the left and pay that price to get rid of the left tail

Nothing more I can say on the subject except that those who have considered and adopted this strategy have been very happy about having done so. And not just through this recent market fall either.

PS-if you go back to 1926 you find that value has been more volatile than TSM--at least that was the case the last time I checked the data (don't have time today to do that). And certainly small has as well. But that is individually as separate asset class--keep in mind these are independent risk factors though and thus one should look at overall portfolio volatility.

[foglifter]

Larry,

I apologize if my question is off-topic. Which one of your books/articles would you recommend as a good source to comprehend your idea of using SV-tilting and increased fixed income part?

Thank you

foglifter

[larryswedroe]

None of them cover that topic

I think too complex for average investor

[Henry Sadovsky]

.

[Henry Sadovsky]

.

The three factor model was not built up from fundamental considerations. A theory has been draped over past return data. The predictions of such a theory (e.g. expected equity portfolio return) must, and will be, tested.

[james22]

[larryswedroe]

henry
What we do is to take our estimate of the ERP using the Gordon Model and then add the expected returns for the funds using the historical risk premiums and the loading factors.

James
there is no right answer to AA. As I have stated in addition to TE one should consider how one's labor capital correlates with the three risk factors as well.

When I was young there was no three factor model so while all equity I was also like most investors today close to a TSM type portfolio.

Note I don't have an ideal portfolio but I do think that for those with stable incomes and low marginal utility of wealth a prudent strategy is a very low equity allocation with a very high tilt to small/value, like mine.

An ISRM would not advocate anything, just educate people about the pros and cons of choices and who should consider tilting to asset classes and who should consider tilting away from them. And issues like that

[Henry Sadovsky]

[wbond]

SmallHi must be on vacation.

[Henry Sadovsky]

.

[wbond]

[larryswedroe]

Henry-all I do is take the div yield add an amount for expected real growth --then add the loading factors.
Each person should do for themselves because it depends on what funds you use.

Larry

[larryswedroe]

Henry

Some further thoughts I hope you find helpful on this issue

Anyone who tilts away from TSM to size and value and then makes the AA (equity to fixed) decision after deciding on the need to take risk is already doing what I am suggesting investors consider --

Consider the investor who decides to invest in TSM and let's assume they expect a 5% real return plus say 1% inflation. That results in a TSM expected return of 6%. If your need to take risk is 6% then you have to be 100% equity. Now on other hand we show clients three model portfolios that are used only as starting points for discussions--we call them Risk Target 1, 2 and 3, with progressively greater tilts to the risk factors of size and value. Each roughly increases expected returns by 1%. So that RT 1 would have expected return of 7, RT2 of 8 and RT 3 of 9. So now you can calculate how much equity exposure you need based on which model you decide to use. And you can look back historically to see what the dispersion of annual returns was and look and see what the annualized returns and SD was to help you figure out which is best for you (also taking into account issues like labor capital, TE).

While the RTs don't match up exactly with DFAs core portfolios you might think of them as RT similar to core 1, RT2 similar to Core 2 and RT 3 between Core 2 and Vector. Thus a Vector fund would have an even lower need to take equity risk. And then of course you can tilt even more and lower your need to take equity risk--taking on different risks.

But anyone adding size or value tilts can lower their equity allocation and still meet their expected return to match their need to take risk. My own personal portfolio is just an extreme version of the idea--more extreme than vector.

Hope that helps.

[wbond]

Larry,

If I understand Henry’s point correctly:

You are measuring risk as SD (volatility).
This may adequately explain the size premium, but not the value premium.
Adhering to EMH, FF prefer to explain the value premium as due to taking risk rather than to behavioral factors.
There is no clear measurement of this risk for the value premium, and it is clearly not the SD.

You then are using the past value returns to predict a future value premium, but are using the SD as your measurement of risk. So, of course, you will have higher returns with the same risk if these are your assumptions. Logically, however, the value premium must either be behavioral (and thus, either risk-free due to persistent behavior, or may disappear due to EMH), or it must carry an additional risk that you are not measuring and are not taking into account.

Note that, for simplicity, the above ignores the argument about the historical costs of owning SV.

I have written this as much to clarify my own understanding as for any other purpose. Please (Henry, Larry, others) correct me if this does not state the position adequately.

(FWIW, I personally tilt fairly aggressively).

(Edited for a comma).

[larryswedroe]

wbond

First, there is this persistent myth that value stocks have not been more volatile than the market. I have already stated that is wrong and yet people keep repeating it. So here are the facts. 1927-2008
returns/SD
TSM 9.5/20.6
FF LV 10.0/27.2
FF SV 13.0/35


And even if you start from 64 the data is similar.


And if you look at the SDs by sub asset class you find
LG 22; LV 27.2, so value more volatile
SG 34; SV 35, so value more volatile
So your statement that it is clearly not SD was clearly wrong.

So if you want to define risk as SD clearly value stocks have higher SD and there is no behavioral story at all. It is a risk story.

Second, I have clearly said over and over again that SD is not RISK, just one form of risk. Skewness and kurtosis can also come into play, and most importantly how the risk of an asset correlates with other assets, including human capital, matter. So that an asset that has the risks show up at the wrong time relative to labor capital risks may carry a high risk premium even if the SD is low. That in fact IMO is the answer to the so-called equity risk premium puzzle (the ERP was too high to be explained by SD).

Third, either the value premium is a risk story (IMO which it mostly is) and then you have the data to guide you, or it is a behavioral story and you have the free lunch (then certainly you should choose my suggested strategy if you believe it).

Fourth, the key why this works IMO is that the risks factors are independent. Correlations are very low. For size factor to market it is just .38 and for price to market it is just .11 and for size to price it is just .04. Thus you are adding unique or independent risk factors with very low correlation, having a positive impact on the volatility of the portfolio (with the reduction also in the beta exposure and volatility of beta is lot higher than volatility of the other factors).

I hope that is helpful

[wbond]

[Henry Sadovsky]

.

[wbond]

Larry,

Again, thanks kindly for your above reply.

Let me preface this by stating that the following questions are offered only in the spirit of better understanding your points, and are not intended to seem argumentative.

In thinking about this further, logically (i.e. not mathematically):

1. If the size/value premium is a risk story only – then how does your portfolio reduce expected long-term risk (of any kind, however defined), assuming the comparison portfolio (TSM, or the like) has some fixed income as well? I say long-term since I realize that there will be significant tracking difference (not just error). In other words, if one has SV-tilted portfolio at a reduced equity % that “should” have the same return and risk (however defined, however measured) as TSM 80%/ST bond 20% (to be simple), where’s the difference outside of the tracking differences if the predictions are correct? The SV-tilted portfolio (with lower equity %) should have exactly the same worst-case scenario losses if the premium is explained completely by risk, shouldn’t it? This should be simple arithmetic, I would think. Clearly we can’t predict the efficient frontier for any future period, but if one aspect of future risk is worst-case bear market losses, where is the flaw in this logic?

2. Under “third” you state that if there is a behavioral story then there is a free lunch – but can’t it go away once understood?

3. Under “fourth” you speak of correlation as the “key to why this works.” But isn’t that substituting the treatment of small and value as asset classes and their effect on a rebalanced portfolio with the question Henry persists with regarding the SV premium itself?

Thanks,

wbond

[larryswedroe]

Wbond
First, I really cannot understand Henry's comments. I don't know how he draws the conclusions he does. What he writes above I don't agree withh and what he says I believe is wrong. Like the first statement---I do believe the value story is mostly a cost of capital story. In other words the value premium is a risk story but the size of the value premium may be larger than it has been if there was no behavioral story at all.

Second, if the value premium is behavioral to any degree either it will persist--and you get th free lunch or it will go away and you are still ahead because you get a one time gain as it is arbed away. So Henry is again wrong in saying I have neglected it. I have explictly talked about this many times. And the part he talks about re value being clobbered, again he is wrong as that is exactly what happened in the Great Depression and it is exactly I have pointed out to both the board and clients. Unfortunately Henry makes statements as if fact but they are not fact.

Third, Markowitz won Nobel Prize by showing you can add risky asset classes without raising risk of portfolio. That is just the same thing I am showing-These are independent risks that have low correlation and the SD of the asset class we are reducing is greater than the SD of the asset classes we are adding. So the combination of those factors leads to a lower dispersion of returns. Clearly no guarantee that it will work. But Henry's point about the value getting crushed IMO works like this:
I cannot see value getting "crushed" unless beta also gets crushed. If that happens I am almost certainly better off because beta will get crushed more and thus my portfolio will have less downside risk. I don't get if value underperforms when beta is positive because I am trying to cut left tail

I hope that helps

Henry-see above. Plus why do you attribute behaviors to people--like Larry agrees but won't state so? Rhetorical question. Nothing more to add.

[Henry Sadovsky]

.

1)

[larryswedroe]

Henry
You simply amaze me. You make statements that are totally incorrect-then you spin them again and make further false statements. So I will now ignore (not read) your posts, So do not expect any responses. Join the Ferri club.

But as my final comments for you.

I have long stated that IMO the value premium is mostly a risk story but there is some evidence on the behavorial side--that is why I have called it not a free lunch but possibly a free stop at the dessert tray.

I do not believe at all that it is an anomaly (except for the possibility of the size of the premium being larger than would have been the case) and thus do not attribute any likelihood that it is only a behavioral story. And thus that it cannot be arbed away. It is a cost of capital story. There are many academic papers on the economic logic of the risk story. Including the issue of the risks correlating with labor capital risks.

BTW-if it is a behavioral story than why have not the behavioral funds created to exploit the behavioral errors been able to generate excess profits? That is the true test of an anomaly. To be an anomaly you haev to show it is exploitable. There is a new paper showing they have failed to do so. So much for that story. I also wrote about several of the bigger behavioral funds a couple years ago and they had failed to outperform simple value benchmarks like DFA funds and I just updated it and found the same thing.

[Henry Sadovsky]

.

[larryswedroe]

Just a correction

The value premium is totally a cost of capital story. The only issue is how much of the incremental premium is risk and how much behavioral. There are many excellent papers on the risk story, clear and common sense risk stories as explanations for the premium. IMO there are some good reasons to believe there is some behavioral story as well--people confusing the familiar with the safe for example. But IMO that is a small part of the story at best. And it is certainly possible it has already been arbed away. But the fact remains that value stocks must have higher expected returns simply because they have higher costs of capital--

The following is one example of the academic papers on the subject; It is from an article I wrote long ago

Since the publication of the study by Eugene F. Fama and Kenneth R. French, “The Cross-section of Expected Stock Returns” in the Journal of Finance in June 1992, financial economists have been trying to discover the source of the value premium. The October 1998 edition of The Journal of Business contains a study, Risk and Return of Value Stocks, by Naifu Cheng and Feng Zhang, that makes the case that value stocks contain a distress (risk) factor. The authors make their case by examining three factors of distress present in value companies:

1) DIV—Firms cutting dividends by at least twenty-five percent.
2) LEV—A high ratio of debt to equity.
3) SEP—A high standard deviation of earnings

The authors found that the three factors all captured the returns information (produced high correlations) contained in portfolios as ranked by book-to-market value. When these three factors were present, returns were greater. Since all three factors have simple intuitive risk interpretations (are associated with firms in distress), they state that it isn’t surprising that the risk factors they studied were highly correlated and were also highly correlated with book-to-market rankings. Their conclusion was that value stocks are cheap because they tend to be firms in distress, with high leverage, and face substantial earnings risk and thus they provide higher returns due to the greater risks facing value investors.

And here is another from that same article I wrote:

A study, “The Value Premium,” by Lu Zhang provides further support to this risk story. His study concluded that the value premium can be explained by the asymmetric risk of value stocks—“they are more risky than growth stocks in bad times and less risky in good times, but to a much lesser extent.”2 Zhang explains that asymmetric risk of value companies exists because value stocks are typically companies with unproductive capital. Asymmetric risk is important because:
·Investment is irreversible—once production capacity is put in place it is very hard to reduce. Value companies carry more non-productive capacity than do growth companies.
·In periods of low economic activity companies with non-productive capacity (value companies) suffer greater negative volatility in earnings because the burden of non-productive capacity increases and they find it more difficult to adjust capacity than do growth companies.
·In periods of high economic activity the previously non-productive assets of value companies become productive while growth companies find it harder to increase capacity.
·In good times capital stock is easily expanded, while in bad times adjusting the level of capital is an extremely difficult task, and is especially so for value companies.

Zhang also observes that:
·Recessions happen with far less frequency than good economic times.
·The longevity of recessions is far shorter than good times.

When these facts are combined with a high aversion to risk by investors (especially when that risk can be expected to show up when their employment prospects are more likely to be in jeopardy) the result is a large and persistent value premium. The authors of another study, “Equilibrium Cross-Section of Returns,” came to the same conclusions as did Zhang.3

Don't know if those links still work, but if not those interested I assume can find the papers. And there are many others.


And we have this, from another article I wrote:

Ralitsa Petkova’s “Do the Fama-French Factors Proxy for Innovations in Predictive Variables?” Journal of Finance (April 2006,) found that value (and small) tend to be firms under distress, with high leverage and high uncertainty of cash flow. Therefore, shocks to the default spread (the spread between bonds of highly rated bonds and lower-rated credits) explains the cross-section of returns and is consistent with value being a measure of distress risk. In addition, growth stocks are high-duration assets (much of their value comes from expected future growth), making them similar to long bonds. Value stocks are low-duration assets, making them more similar to short-term bonds. Thus, shocks to the term spread (the difference between short-term bonds and long-term bonds) also explains the cross-section of returns and is also consistent with value being a measure of distress risk.
So, first we have the claim that value is anomaly because it has higher returns but not higher SD. I showed that is false. Value does in fact have higher SD. So then people claim well what is the source of the extra risk. And there are plenty of papers explaining the sources.
Thus, the responses are not my opinions but the findings from academic papers. Now if someone wants to claim the findings are wrong, fine.

2.Lu Zhang, “The Value Premium.” January 2002, http://assets.wharton.upenn.edu/~zhanglu/
3.Joao Gomes, Leonid Kogan, and Lu Zhang, “ Equilibrium Cross-Section of Returns, March 2001. http://assets.wharton.upenn.edu/~zhanglu/

[tomser]

LARRY SWEDROE

this may be off topic but I really hope you can help retirees in bogleheads

In general term, what would you recommend ( in your opinion ) fixed in come portion of allocation ( 30% stock / 70% bond -,low risk-capital preservation ) and which
bond funds ( vg bond funds or individual bonds ) in this melting down economics Let say-retired 25 to 29 % tax bracket- taxable and tax differed both cases.

examples;

vg tip
vg REIT
vg munis-intermediate -long term -or California
vg GNMA
vg treasury intermediate or long term
vg high yield
vg total bond
vg short term bond index or intermediate bond index
vg money market or CDs
vg short term invest grade or intermediate invest grade
and what % allocation ???

Appreciate your input

tomser

[StephenT]

[larryswedroe]

Tomser
First so you can make your own informed decision I suggest you read The Only Guide to a Winning Bond Strategy You'll Ever Need.

Also for more information on many of the fixed income options suggest you read the Alternative Investments Only Guide.

As to your question specifically
For tax advantaged accounts I would be all TIPS in most cases--unless risk premium for unexpected inflation was very high. And right now that is certainly not the case. That is what the academic literature finds.

For taxable I would be in munis of the highest grade. Well diversified by credit.

You don't need anything other than that. If you want/need more expected return take the risks on the equity side of the portfolio where you can diversify them more effectively and earn the return in more tax efficient manner.

Hope that helps

[Sammy_M]

Larry and Henry (and in other threads...Rick, SmallHi, etc.),

I imagine these exchanges are trying at times, but I wanted to say thank you to all for having it. Debate brings out info and provokes thought in a way like no other. I personally have found it quite informative and helpful.

It exemplifies how this forum offers value beyond that which can be obtained by just working with an advisor. I might challenge an advisor several times about the persistence of the S/V premium and then I'd probably get to a point where I was satisified. And then 5-10 years may pass where S/V underperforms and I'd say to myself "maybe my advisor was just wrong" and I'd move back to TSM. With the type of vigorous debate and challenging that happens on BH, and the supporting evidence (for both sides) that comes out, it can strengthen one's resolve around adopting/embracing a particular strategy.

[larryswedroe]

Sammy

It is interesting to note the following

A) At least the last time I looked at the data, the value premium was more persistent than the equity premium, over all the time frames I looked at (i.e., one year, five year, ten year, etc). Yet while no one questions the ERP so many question the value premium.

B) People claim the value premium is an anomaly because the SD of value is lower than TSM, but that is not true as I have showed, yet the same people keep repeating that story

C) there are many academic papers showing the existence of simple, logic risk based explanations for the value story yet they get ignored by the skeptics.

D) When FF "discovered" the value premium the skeptics said it was data mining--it would not hold up to out of sample tests. But the value premium has not only held up since the paper was written but FF went back and reconstructed the data back to 1926 and the premium was there too, and about the same size. And then there have been many studies on the value premium in out of sample tests in overseas markets, both developed and EM and amazingly it shows up there and in about the same size--pure coincidence?

So you look at the data and make your own decision.

Best

[Henry Sadovsky]

.

[larryswedroe]

http://www.indexuniverse.com/s....h/920.html

[Henry Sadovsky]

.

To Whom It May Concern,

Studies supporting the common sense notion that small size and high book-to-market firms offer the investor greater risk (and thus must bear a higher cost of capital) are of very little interest if they do not also offer a theory and methodology for estimating what that higher cost of capital 'should' be. A FF portfolio of Small-Value firms 'should' have a higher cost of capital than the market portfolio. But... how much higher is warranted? Is 5%/yr (annualized expected return) above that of the market portfolio reasonable?

The only study Larry has so far (AFAICT) cited that gets to this question is the one of Gomes, Kogan, and Lu Zhang (see above posts). It appears to me that for a 100% FF SV equity portfolio, they predict a return of approx. 2.5%/yr (gross of additional expenses) above that of the market portfolio.

Anyway... that'll be it from me for now as something unexpected has occurred. I have (for the time being) become bored with this topic.



Henry
_________________
"What we can't say we can't say, and we can't whistle it either."
Frank P. Ramsey"

(f.k.a. Zalzel)

[james22]

[larryswedroe]

James
Few quick thoughts

I think basically you have it

For those with all fixed income in tax advantaged than TIPS are all you really need, especially if want to keep it simple

I chose the portfolio I did because I thought the all SV portfolio to extreme for many if not most people--far less likely to have discipline and patience that I have---need to really understand this stuff so you will stay the course

For same reason I did not include CCF--first simplicity and second most don't understand it and should not invest in anything you don't understand

[Rodc]

Larry,

If you still have time and energy, there is something I don't understand about the cost of capital story. I understand, I think, the ideas linking cost of capital to expected returns. What I don't understand is how it explains the value risk premium. That is, how does it explain the return above and beyond the basic beta return?

Does it go, more or less, like

1) risk premium = beta*(Rm-Rf) + gamma*(CCm-CC) where the second term is the cost of capital above the market average cost of capital, and beta is independent of cost of capital,

or is it just

2) risk premium = cost of capital, no reference to beta risk premium,

or

3) something else?

If 2, why expect cost of capital for a value company to be higher than beta*(Rm-Rf)?

TIA.
_________________
"all standard caveats apply"

[james22]

Thank you, Larry.
_________________
Please assume my post refers to my Bogle-approved 15% Tactical Asset Allocation or 5% Funny Money.

[richard]

Rod,

The multifactor models were developed because CAPM (single factor beta model) did not do a great job of explaining returns for small and value stocks.

The cost of capital story is really just a way of saying that risky stocks have a higher return. Investors demand a higher return before investing in a riskier security. Cost of capital is the conclusion of a model (cost of capital is the same as expected return), not an input to the model.

[Rodc]

[StephenT]

Rod,
It has been decades since I took econometrics and statistics classes but I thought multiple regression captured linear factors/independent variables (and non-linear factors through transformations) explaining the dependent variable. Larry and others will correct me if I am wrong but isn’t that the essence of what FF did with small (SMB) and value (HML). They essentially refined the equity component of the cost of capital. {The cost of capital reflects both the cost of equity and the cost of debt.} So a company’s cost of equity is influenced by not only its riskiness relative to the entire market, beta, but also its exposure to other factors. Hopefully that helps and I am sure I will learn from the responses.

[Rodc]

[richard]