Let me get this straight about SCV & factor diversification

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Let me get this straight about SCV & factor diversification

Post by Browser »

There's been much discussion about factor investing, which is advocated by Larry Swedroe. In many comments, he has argued that one should not put all one's eggs in one factor basket, but diversify across factors: beta, value, size, and momentum. Am I correct that he advocates a high tilt toward SCV because SCV is actually more diversified across beta, value, and size than is TSM? In other words SCV, even though it represents just a small slice of the total market (and would therefore seem to be undiversified) is actually more diversified than the total market from a factor perspective? From a factor perspective SCV represents superior diversification and therefore is superior to holding TSM?
We don't know where we are, or where we're going -- but we're making good time.
User avatar
Clearly_Irrational
Posts: 3087
Joined: Thu Oct 13, 2011 3:43 pm

Re: Let me get this straight about SCV & factor diversificat

Post by Clearly_Irrational »

Larry will have to speak for himself about what he advocates.

If you accept that a multi-factor model more accurately describes the returns of stocks and bonds then the decision about what to own becomes more complicated than just your stock/bond split. Beta remains the largest factor and so still has a heavy impact, but now you need to express preferences along the size, value, momentum, term and credit dimensions as well. The whole point of the multi-factor model was to show that many of those things weren't a free lunch. I don't think you can argue that SCV is objectively better than TSM in a multi-factor model, just that it expresses a different set of risk/return preferences.
countmein
Posts: 653
Joined: Fri Dec 06, 2013 8:10 pm

Re: Let me get this straight about SCV & factor diversificat

Post by countmein »

That is correct, SCV more diversified from a factor perspective. That said, as I commented in your other thread, I still wouldn't be comfortable with only handful of small or micro stocks, but that is just me. If you want to go full tilt you have to really believe in it. I happen to believe with more certainty that small value will get absolutely obliterated in deflationary times more than I believe that it will outperform by any wide margin in 'normal times'.
dbr
Posts: 46137
Joined: Sun Mar 04, 2007 8:50 am

Re: Let me get this straight about SCV & factor diversificat

Post by dbr »

It is first necessary to be clear what the word "diversification" means in the various contexts in which it is used. It helps immensely to look at what properties one might want in a portfolio, to quantify those properties, and then to show how different styles of portfolios behave regarding the properties of interest. The latter is an extraordinarily difficult task which is never really completely accomplished satisfactorily. If your definition of diversification is that a you have significant loading on value and size risk factors in addition to market beta, then a portfolio significantly tilted to SCV is more diversified, by definition, than TSM. Why such a choice would be superior takes more explanation than simply the statement that it is "more diversified," which is a circular argument. None of this objecting constitutes an argument that factor diversification is not a good idea. Note that "risk factor" means nothing more than "that which predicts return."

Probably a less negative comment about the whole thing is that the language used here, "diversification," is a short hand reference to a large body of analysis and recommendation and should not be thought of in this case as invoking an obvious and well known fundamental that we all know is good to have. Investment by invocation is a bad idea always and everywhere.
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Let me get this straight about SCV & factor diversificat

Post by larryswedroe »

Browser
To be clear I have never advocated ANY particular portfolio. I only recommend that people consider such a strategy, but the right answer depends on the unique situation and even personality of the investor
And one certainly should never invest in anything they don't fully understand
I would add that while most of my firm's clients have highly tilted portfolios, maybe 10% have a Larry type portfolio and they tend to be either very young and then use it with very high if not 100% equity allocation or very wealth and use it to cut tail risks.
Larry
Topic Author
Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Re: Let me get this straight about SCV & factor diversificat

Post by Browser »

larryswedroe wrote:Browser
To be clear I have never advocated ANY particular portfolio. I only recommend that people consider such a strategy, but the right answer depends on the unique situation and even personality of the investor
And one certainly should never invest in anything they don't fully understand
I would add that while most of my firm's clients have highly tilted portfolios, maybe 10% have a Larry type portfolio and they tend to be either very young and then use it with very high if not 100% equity allocation or very wealth and use it to cut tail risks.
Larry
Not saying that you are peddling SCV. I am trying to more fully understand the basic rationale for considering SCV. Is it essentially that it represents better factor diversification than the total market (across beta, value, and size). Consequently, If I as an investor am sufficiently persuaded that the multi-factor perspective is correct, then it would follow that SCV (considering only risk/return) is a superior investment to TSM, even though it represents only a small corner of the total market?
We don't know where we are, or where we're going -- but we're making good time.
User avatar
SimpleGift
Posts: 4477
Joined: Tue Feb 08, 2011 2:45 pm

Re: Let me get this straight about SCV & factor diversificat

Post by SimpleGift »

Browser, not sure exactly what you are asking in this thread, but my sense is that each of the risk factors has independent (and changing!) correlations with the overall market and with each other (chart below). It's hard to say that the size and value factors are better than TSM or any other risk factor — they just behave differently and can add diversification to TSM. Hope this helps.

Image
Source: R-bloggers

Note: Shown are the Fama-French "industrial strength" factors and most SCV funds would have higher correlations to market.
Last edited by SimpleGift on Thu Sep 04, 2014 9:44 pm, edited 1 time in total.
User avatar
Blue
Posts: 1178
Joined: Sat Jul 12, 2008 10:18 pm

Re: Let me get this straight about SCV & factor diversificat

Post by Blue »

countmein wrote:. I happen to believe with more certainty that small value will get absolutely obliterated in deflationary times more than I believe that it will outperform by any wide margin in 'normal times'.
Curious why you believe this?
steve_14
Posts: 1507
Joined: Wed Jun 20, 2012 12:05 am

Re: Let me get this straight about SCV & factor diversificat

Post by steve_14 »

Browser wrote:There's been much discussion about factor investing, which is advocated by Larry Swedroe. In many comments, he has argued that one should not put all one's eggs in one factor basket, but diversify across factors: beta, value, size, and momentum.
Yep, that's why I only invest in stocks based in Brooklyn. I get the unique risk factors of America, New York State, NYC, and Brooklyn itself. I'm clearly more diversified than a TSM portfolio, which gets only one of these four factors.
Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: Let me get this straight about SCV & factor diversificat

Post by Rodc »

steve_14 wrote:
Browser wrote:There's been much discussion about factor investing, which is advocated by Larry Swedroe. In many comments, he has argued that one should not put all one's eggs in one factor basket, but diversify across factors: beta, value, size, and momentum.
Yep, that's why I only invest in stocks based in Brooklyn. I get the unique risk factors of America, New York State, NYC, and Brooklyn itself. I'm clearly more diversified than a TSM portfolio, which gets only one of these four factors.
Cute.

But mocking is not exactly an intelligent argument.

Do you actually have something useful to add to the conversation?
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.
dbr
Posts: 46137
Joined: Sun Mar 04, 2007 8:50 am

Re: Let me get this straight about SCV & factor diversificat

Post by dbr »

Browser wrote:
larryswedroe wrote:Browser
To be clear I have never advocated ANY particular portfolio. I only recommend that people consider such a strategy, but the right answer depends on the unique situation and even personality of the investor
And one certainly should never invest in anything they don't fully understand
I would add that while most of my firm's clients have highly tilted portfolios, maybe 10% have a Larry type portfolio and they tend to be either very young and then use it with very high if not 100% equity allocation or very wealth and use it to cut tail risks.
Larry
Not saying that you are peddling SCV. I am trying to more fully understand the basic rationale for considering SCV. Is it essentially that it represents better factor diversification than the total market (across beta, value, and size). Consequently, If I as an investor am sufficiently persuaded that the multi-factor perspective is correct, then it would follow that SCV (considering only risk/return) is a superior investment to TSM, even though it represents only a small corner of the total market?
To cut to the chase this depends on what you mean by superior. FF shows that it can be possible to increase the return of a portfolio over TSM by adding loadings on small and value factors. For some people this would be a definition of superior, the "my portfolio will beat your portfolio" school of thought. What we are talking about here is, of course, that illusory statistic "expected return" with all it does and doesn't imply. A little more astute questioner would ask what other properties of the portfolio are changed. For example as the expected return increases does the standard deviation of expected return increase, decrease, or remain the same. In the latter two cases the argument for superiority gets better, but in the first case the situation is not so clear. Maybe if one can calculate an expected Sharpe ratio and that gets better, then the new portfolio is superior. A question one could ask is whether the safe withdrawal rate in retirement is higher for a heavily tilted portfolio than for TSM, over various bond allocations being equal. One could also ask if there is an advantage over various bond allocations not being equal. This might be the "Larry" portfolio, heavily tilted to SCV but risk diluted with less allocated altogether to stocks hoping to reduce extreme downside risks with expected return and standard deviation of returns not much helped. These are the things that one could mean by how does the portfolio behave and which of those things does one want.
User avatar
Kevin M
Posts: 15750
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Let me get this straight about SCV & factor diversificat

Post by Kevin M »

Browser: Maybe this will help:

Image

Note that the loading on the market factor is close to 1. Although the market beta here isn't the same as the CAPM market beta, since the regression is done on more than just the market factor, it still indicates that much of the SCV fund return is determined by the overall US stock market. In addition, you get the additional exposure to the size and value factors.

If you run the regression on TSM, you'll see the loading on the market factor is 1, and the loading on the size and value factors is close to 0. Here is a link to the regression for both funds that shows this: Fama-French Factor Regression Analysis for VTSMX and VISVX.

So I think this tends to provide support that the answer to the question in your OP is "yes". I believe the small-cap-value fund is well diversified across companies and sectors, diversifying away much of the unsystematic risk, and leaving you exposed to the supposedly systematic risk factors of market, small, and value. Note though that it leaves you exposed to country-specific (US) risk, which I assume is why Larry typically suggests considering including international small and value stocks as well.

Kevin
If I make a calculation error, #Cruncher probably will let me know.
thx1138
Posts: 1164
Joined: Fri Jul 12, 2013 2:14 pm

Re: Let me get this straight about SCV & factor diversificat

Post by thx1138 »

Deleted, not useful
Last edited by thx1138 on Thu Sep 04, 2014 9:38 pm, edited 1 time in total.
dbr
Posts: 46137
Joined: Sun Mar 04, 2007 8:50 am

Re: Let me get this straight about SCV & factor diversificat

Post by dbr »

Kevin M wrote: So I think this tends to provide support that the answer to the question in your OP is "yes". I believe the small-cap-value fund is well diversified across companies and sectors, diversifying away much of the unsystematic risk, and leaving you exposed to the supposedly systematic risk factors of market, small, and value. Note though that it leaves you exposed to country-specific (US) risk, which I assume is why Larry typically suggests considering including international small and value stocks as well.

Kevin
Kevin, I think your post is very helpful for explaining what one is saying when one says one has factor diversification. I also like that you bother to bring in the only "pure" use of the word, which is diversification of unsystematic or diversifiable risk.

I am less enthusiastic about the rest. Excepting the FF return model being "exposed" to risk factors is meaningless term beyond the circular definition that it means the regression found significant regression coefficients on the size and value factors. The meaning in the FF model is that we have a fairly robust result that says we can explain expected returns in terms of these factors and that more heavily loaded portfolios would have higher expected return. Aside from that, what have we learned that would be important to an investor who is thinking that tilting to SCV would be a good idea?
User avatar
Rick Ferri
Posts: 9703
Joined: Mon Feb 26, 2007 10:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

Re: Let me get this straight about SCV & factor diversificat

Post by Rick Ferri »

There's more to factor investing than meets the eye.

1) it is always more expensive than a simple TSM (beta) portfolio.
2) it adds unfamiliar risks to a portfolio that creates tracking error.
3) when the tracking error is large and negative, there is the added risk of capitulation.

Here is how I recommend dealing with these issues:

1) don't factor invest just because everyone is talking about "smart beta"
2) don't factor invest unless you fully understand the risk and cost
3) don't factor invest unless you're committed to it for life
4) do try to find the cheapest factor products for the price (lowest cost per unit of risk)
5) do track your factor weights and rebalance on occasion
6) do try to squeeze as many factors into one fund as possible (DFA is good at this)
7) only put a reasonable % into factor funds - I recommend no more than 25%

Hope that helps!

Rick Ferri
Last edited by Rick Ferri on Thu Sep 04, 2014 7:42 pm, edited 1 time in total.
The Education of an Index Investor: born in darkness, finds indexing enlightenment, overcomplicates everything, embraces simplicity.
dbr
Posts: 46137
Joined: Sun Mar 04, 2007 8:50 am

Re: Let me get this straight about SCV & factor diversificat

Post by dbr »

Here is an article that displays more information about how factors, returns, and standard deviations of returns work out:

http://seekingalpha.com/article/2035813 ... ctor-model
golfvestor
Posts: 80
Joined: Fri Aug 22, 2014 11:50 pm

Re: Let me get this straight about SCV & factor diversificat

Post by golfvestor »

Almost any small group of stocks will be more diversified in terms of the number of factors because of the way factors are defined. Factors, except for beta, are defined such that the market always has a zero loading, so it's almost a tautology. SCV might out-perform the total market. It might even do so on a risk adjusted basis, under some definitions of risk. However, I don't think you can claim a small basket of stocks is more diversified than owning the total market, except by using a definition of diversification that is antithetical to its normal and customary definition.
lee1026
Posts: 396
Joined: Sat Jun 21, 2014 7:47 pm

Re: Let me get this straight about SCV & factor diversificat

Post by lee1026 »

I am less enthusiastic about the rest. Excepting the FF return model being "exposed" to risk factors is meaningless term beyond the circular definition that it means the regression found significant regression coefficients on the size and value factors. The meaning in the FF model is that we have a fairly robust result that says we can explain expected returns in terms of these factors and that more heavily loaded portfolios would have higher expected return. Aside from that, what have we learned that would be important to an investor who is thinking that tilting to SCV would be a good idea?
Assuming the factors are real and the premiums are persistent, we can imagine building the following funds:

1. Pure Beta. TSM is a good enough proxy for this.
2. Pure Small. Go long on a small ETF and then short fund(1) so that it have a beta of 0.
3. Pure Value. Go long on a value ETF and then short fund(1) and fund(2) so that it have a beta of 0 and size load of 0.
Repeat the procedure for other factors that you believe also exist.

At this point, we have 3 funds that don't correlate with each other at all (by definition!) and all of whom have large positive expected value. Build a mixture of the 3 based on your estimates for the size of the factors. Assuming that we are dealing premiums that are as large as is often claimed here, the ideal mixture will probably be a mix of a small fund, a value fund, and shorting the S&P e-mini. If all of the factors that are often claimed here are real, you should be able to construct a fund that have an expected returns that is far larger than the standard deviation. At that point, leverage it up to taste and reap speculator returns.
longinvest
Posts: 5672
Joined: Sat Aug 11, 2012 8:44 am

Re: Let me get this straight about SCV & factor diversificat

Post by longinvest »

golfvestor wrote:Almost any small group of stocks will be more diversified in terms of the number of factors because of the way factors are defined. Factors, except for beta, are defined such that the market always has a zero loading, so it's almost a tautology. SCV might out-perform the total market. It might even do so on a risk adjusted basis, under some definitions of risk. However, I don't think you can claim a small basket of stocks is more diversified than owning the total market, except by using a definition of diversification that is antithetical to its normal and customary definition.
But that wouldn't sell DFA and AQR funds! :mrgreen:

When I look at the proportion of the market that small-cap value (SCV) represents, and I think about the mass media coverage about the superiority of SCV, I get the feeling that the demand for SCV is superior to the offer. Basic economic theory says that this should drive up the price (thus reduce returns). But, what do I know about it?

As I don't understand factors well enough, I don't invest in them. I like to keep things simple. :happy
Variable Percentage Withdrawal (bogleheads.org/wiki/VPW) | One-Fund Portfolio (bogleheads.org/forum/viewtopic.php?t=287967)
MrMatt2532
Posts: 454
Joined: Sun Mar 15, 2009 11:58 am

Re: Let me get this straight about SCV & factor diversificat

Post by MrMatt2532 »

lee1026 wrote: At this point, we have 3 funds that don't correlate with each other at all (by definition!) and all of whom have large positive expected value.
I'm ok with most that you wrote, however the correlations are not zero and there is the nothing in how they are defined that would imply this. See the figure in the post by Simplegift above!
Day9
Posts: 1000
Joined: Mon Jun 11, 2012 6:22 pm

Re: Let me get this straight about SCV & factor diversificat

Post by Day9 »

golfvestor wrote:Almost any small group of stocks will be more diversified in terms of the number of factors because of the way factors are defined. Factors, except for beta, are defined such that the market always has a zero loading, so it's almost a tautology. SCV might out-perform the total market. It might even do so on a risk adjusted basis, under some definitions of risk. However, I don't think you can claim a small basket of stocks is more diversified than owning the total market, except by using a definition of diversification that is antithetical to its normal and customary definition.
Most Bogleheads who own SCV funds do so in addition to TSM funds. Tilting like this gets you closer to an equal weight portfolio which would appear to be intuitively more diversified than the 2% Apple, 2% Exxon, etc purely cap weighted portfolio
I'm just a fan of the person I got my user name from
lee1026
Posts: 396
Joined: Sat Jun 21, 2014 7:47 pm

Re: Let me get this straight about SCV & factor diversificat

Post by lee1026 »

I'm ok with most that you wrote, however the correlations are not zero and there is the nothing in how they are defined that would imply this. See the figure in the post by Simplegift above!
You can construct them to be zero by longing or shorting the other factors. E.g. if Value is correlated negatively against MOM, you long value in your MOM fund to make it neutral.
User avatar
ogd
Posts: 4876
Joined: Thu Jun 14, 2012 11:43 pm

Re: Let me get this straight about SCV & factor diversificat

Post by ogd »

I think we should not use the word "diversification" for this concept.

Call it "coverage" or "exposure" or something like that. Take "coverage" -- it's a fairly strong word that implies completeness. Nothing wrong with it.

But "diversification", when what you're doing is concentrating in certain types of companies observed to do well in the past, is just a little too Orwellian. Even if those past observations are rock solid. The fact remains, fewer companies or (sometimes) sectors have to fail for you to get really hurt, and fewer have to wildly succeed for you to really fall behind.
MrMatt2532
Posts: 454
Joined: Sun Mar 15, 2009 11:58 am

Re: Let me get this straight about SCV & factor diversificat

Post by MrMatt2532 »

Day9 wrote: Most Bogleheads who own SCV funds do so in addition to TSM funds. Tilting like this gets you closer to an equal weight portfolio which would appear to be intuitively more diversified than the 2% Apple, 2% Exxon, etc purely cap weighted portfolio
I've tilted since day one myself, however, it has nothing to do with your reasoning. Take a simple thought experiment: You have a fictitious world economy with 3 companies and each is worth 1 megabuck. By either cap weighting or equal weighting you should hold 1/3 of each company. Now, let's say one of the companies buys out one of the others. Now, what should the investor do? The cap weighted investor wouldn't have to do anything, however the equal weighted investor needs to sell part of the now larger company to buy the smaller company. I would argue doing so is a very arbitrary thing to do. Look at big megacorp companies. You should hold more of those companies because they are literally like many smaller companies considering the extent of their business activities. Cap weighting is a very natural thing to do.

Now, the reason I tilt towards SV is because I believe it has unique risks that lead to increased risk-adjusted returns in comparison with just holding the market.
MrMatt2532
Posts: 454
Joined: Sun Mar 15, 2009 11:58 am

Re: Let me get this straight about SCV & factor diversificat

Post by MrMatt2532 »

lee1026 wrote:
I'm ok with most that you wrote, however the correlations are not zero and there is the nothing in how they are defined that would imply this. See the figure in the post by Simplegift above!
You can construct them to be zero by longing or shorting the other factors. E.g. if Value is correlated negatively against MOM, you long value in your MOM fund to make it neutral.
I'm not sure about that. What you described in your original post was how to construct three funds: 1) market fund, 2) small exposure only fund, and 3) value exposure only fund. Those three funds are not (necessarily) uncorrelated...
User avatar
Kevin M
Posts: 15750
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Let me get this straight about SCV & factor diversificat

Post by Kevin M »

Browser wrote:From a factor perspective SCV represents superior diversification and therefore is superior to holding TSM?
Just to clarify, the purpose of my post was to answer the other questions in the OP, not this one. I didn't mean to say or imply that the loading of SCV on the additional risk factors results in it being superior to TSM. I was just trying to help clarify what is meant by "factor diversification", for example as used by Larry.

Kevin
If I make a calculation error, #Cruncher probably will let me know.
lee1026
Posts: 396
Joined: Sat Jun 21, 2014 7:47 pm

Re: Let me get this straight about SCV & factor diversificat

Post by lee1026 »

I'm not sure about that. What you described in your original post was how to construct three funds: 1) market fund, 2) small exposure only fund, and 3) value exposure only fund. Those three funds are not (necessarily) uncorrelated...
We can define the funds as "the premium that a particular factor brings after accounting for all of the other ones". A factor that does not have above zero returns after adjusting for another factors is a pointless waste of time.
MrMatt2532
Posts: 454
Joined: Sun Mar 15, 2009 11:58 am

Re: Let me get this straight about SCV & factor diversificat

Post by MrMatt2532 »

lee1026 wrote:
I'm not sure about that. What you described in your original post was how to construct three funds: 1) market fund, 2) small exposure only fund, and 3) value exposure only fund. Those three funds are not (necessarily) uncorrelated...
We can define the funds as "the premium that a particular factor brings after accounting for all of the other ones". A factor that does not have above zero returns after adjusting for another factors is a pointless waste of time.
Who said anything about returns not being above zero?
Again:
Fund 1: b3=1, bv=0, bs=0 --> positive expected return
Fund 2: b3=0, bv=1, bs=0 --> positive expected return
Fund 3: b3=0, bv=0, bs=1 --> positive expected return
These funds all have expected positive return and have a correlation between each other that is not zero...
lee1026
Posts: 396
Joined: Sat Jun 21, 2014 7:47 pm

Re: Let me get this straight about SCV & factor diversificat

Post by lee1026 »

No -

Fund 1 - b3 = 1 -> positive expected return
Fund 2 - b3 = -1 bv = 1 -> positive if the value premium is real
Fund 3 - b3 = -0.9 bv = -0.1 bs =1 -> positive if the small premium is real after account for Beta and value.

If the premiums are real, you would be able to construct three funds that are independent and have positive return. Remember, there is no rule that says that you can't short things.
MrMatt2532
Posts: 454
Joined: Sun Mar 15, 2009 11:58 am

Re: Let me get this straight about SCV & factor diversificat

Post by MrMatt2532 »

lee1026 wrote:No -

Fund 1 - b3 = 1 -> positive expected return
Fund 2 - b3 = -1 bv = 1 -> positive if the value premium is real
Fund 3 - b3 = -0.9 bv = -0.1 bs =1 -> positive if the small premium is real after account for Beta and value.

If the premiums are real, you would be able to construct three funds that are independent and have positive return. Remember, there is no rule that says that you can't short things.
???
I'm not debating with you rather the premium is real or not...I'm just simply stating that the 3 funds you described (or that I described) do not have zero correlation with each other. Also, you do realize that a typical small fund has b3~=1 and bs~=0.5 or so and a typical value fund has b3~=1 and bv~=0.5 or so, right?

Btw, I have no problem with this statement:
lee1026 wrote:Remember, there is no rule that says that you can't short things.
That's why I wrote in my original reply to you that what you said makes sense, EXCEPT that the correlations between the funds you described are not zero, therefore not independent.
YDNAL
Posts: 13774
Joined: Tue Apr 10, 2007 4:04 pm
Location: Biscayne Bay

Re: Let me get this straight about SCV & factor diversificat

Post by YDNAL »

Browser » Thu Sep 04, 2014 4:25 pm wrote:There's been much discussion about factor investing, which is advocated by Larry Swedroe. In many comments, he has argued that one should not put all one's eggs in one factor basket, but diversify invest across factors: beta, value, size, and momentum. Am I correct that he advocates a high tilt toward SCV because SCV is actually more diversified exposed across beta, value, and size than is TSM? In other words SCV, even though it represents just a small slice of the total market (and would therefore seem to be undiversified concentrated) is actually more diversified less concentrated than the total market from a factor perspective? From a factor perspective SCV represents superior diversification exposure to factors. and therefore is superior to holding TSM?
No to all questions as stated.

What I see is *misuse* of words from their most basic and widely used definition. In fact, since you mentioned Larry Swedroe, he has attempted to defend that misuse.

IF it is OK with you - to better show what I mean - I felt it necessary to change words in your original post. If it is not OK with you, please let me know and I'll erase this entire darn post. Concentration on a small basket of stocks - "small" relative to ALL investable stocks - is not diversification.

[edit to add link]
Last edited by YDNAL on Fri Sep 05, 2014 7:34 am, edited 1 time in total.
Landy | Be yourself, everyone else is already taken -- Oscar Wilde
longinvest
Posts: 5672
Joined: Sat Aug 11, 2012 8:44 am

Re: Let me get this straight about SCV & factor diversificat

Post by longinvest »

YDNAL wrote:
Browser » Thu Sep 04, 2014 4:25 pm wrote:There's been much discussion about factor investing, which is advocated by Larry Swedroe. In many comments, he has argued that one should not put all one's eggs in one factor basket, but diversify invest across factors: beta, value, size, and momentum. Am I correct that he advocates a high tilt toward SCV because SCV is actually more diversified exposed across beta, value, and size than is TSM? In other words SCV, even though it represents just a small slice of the total market (and would therefore seem to be undiversified concentrated) is actually more diversified less concentrated than the total market from a factor perspective? From a factor perspective SCV represents superior diversification exposure to factors. and therefore is superior to holding TSM?
No to all questions as stated.

What I see is *misuse* of words from their most basic and widely used definition. IF it is OK with you - to better show what I mean - I felt it necessary to change words in your original post. If it is not OK with you, please let me know and I'll erase this entire darn post. Concentration on a small basket of stocks - "small" relative to ALL investable stocks - is not diversification.
The reworded version better matches my intuitive understanding. Thanks.
Variable Percentage Withdrawal (bogleheads.org/wiki/VPW) | One-Fund Portfolio (bogleheads.org/forum/viewtopic.php?t=287967)
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Let me get this straight about SCV & factor diversificat

Post by larryswedroe »

To me this is quite simple
1) Since we know that exposure to factors explains almost all returns, then you should construct a portfolio that gives you the exposure to the factors you want in the amounts you want.
2) Using asset classes is "primitive" way of doing it because you can have say three small value funds that are of the same asset class thus many would think are the same, but have very different loadings on factors.
3) And since factors while explaining the vast majority of differences in returns between diversified portfolios have very low correlations it seems logical to seek to diversify across factors---assuming you want exposure to them.

There is nothing wrong with TSM if it gives you the exposure to the factors you want. And of course it's low cost and tax efficient. But you must recognize that while you do own small and value stocks you have no exposure to these factors because the large and growth stocks you own give you exactly offsetting negative exposure to those factors.

So it depends on how you want to look at diversification. If it's by number of stocks, clearly TSM is the most diversified. But with portfolios that focus on factors you own enough stocks to basically eliminate the idiosyncratic risks of individual companies anyway.

clearly, historically a portfolio that has high exposure to factors and then used high quality bonds has produced superior risk-adjusted returns as well as cutting tail risks. That's fact, not opinion. Whether it will continue to do so in future is another question, and each person has to decide for themselves if there is sufficient reason to believe that will be the case. Obviously I do.

The key as Rick noted is discipline to stay the course. But that's true of TSM as well. TSM lost 0.3% a year from 2000-09 while DFA SV returned 9.1, an underperformance in terms of total return. Each dollar invested in TSM (even before any expense) "grew" to just 97 cents while each dollar in DFSVX grew to $2.40. How disciplined might someone stay losing money for 10 years while observing this growth in small value?
Even worse is the 2000-02 period when TSM lost 14.5 percent per year while DFSVX earned 6.3, growth of dollar was down to 67 cents for TSM while DFSVX was up to $1.21
Now of course reverse was true in late 90s. But not for ten years.

So this discipline thing works both ways, not just one.

Larry
dbr
Posts: 46137
Joined: Sun Mar 04, 2007 8:50 am

Re: Let me get this straight about SCV & factor diversificat

Post by dbr »

larryswedroe wrote:
clearly, historically a portfolio that has high exposure to factors and then used high quality bonds has produced superior risk-adjusted returns as well as cutting tail risks. That's fact, not opinion. Whether it will continue to do so in future is another question, and each person has to decide for themselves if there is sufficient reason to believe that will be the case. Obviously I do.

Thank you. The point is somebody actually has to say this and have the data on which the statement is based. It would probably also be a good idea to add that "risk-adjusted" means risk as defined by volatility of returns, more specifically standard deviation of annual returns. It would also not hurt to explain how one calculates a risk-adjusted return and what the significance of risk adjusted return might be to an investor. I realize this takes a book and not a posting, but that is why I keep calling this conversation a reference to a larger discussion and not something one can explore in a thread on a forum.

Disclaimer: I have no quarrel with the data or even the advice. I just want a spade called a spade instead of hidden behind jargon.


The key as Rick noted is discipline to stay the course. But that's true of TSM as well. TSM lost 0.3% a year from 2000-09 while DFA SV returned 9.1, an underperformance in terms of total return. Each dollar invested in TSM (even before any expense) "grew" to just 97 cents while each dollar in DFSVX grew to $2.40. How disciplined might someone stay losing money for 10 years while observing this growth in small value?
Even worse is the 2000-02 period when TSM lost 14.5 percent per year while DFSVX earned 6.3, growth of dollar was down to 67 cents for TSM while DFSVX was up to $1.21
Now of course reverse was true in late 90s. But not for ten years.

So this discipline thing works both ways, not just one.

Thank you again. This "tracking error" thing is confusing language. Commitment to any particular portfolio design means that comparison to some other portfolio design can result in disappointment for long periods of time. The same thing happens when we allocate an amount to bonds rather than stocks or vice-versa. It is another example of using jargon when plain language would have done better. There is already a usage of tracking error to refer to the failure of an index fund to match the behavior of it's index. There are also more common uses of the word diversification that are contradictory to references to concentrating assets differently than found in the cap weighted total market.

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

Re: Let me get this straight about SCV & factor diversificat

Post by Browser »

Thank you Larry for your clarification. In particular, I'm interested in this comment:
There is nothing wrong with TSM if it gives you the exposure to the factors you want. And of course it's low cost and tax efficient. But you must recognize that while you do own small and value stocks you have no exposure to these factors because the large and growth stocks you own give you exactly offsetting negative exposure to those factors.
If I'm correctly understanding you, an investor who wants exposure to small and value factors will actually not have the exposure he/she might think by holding TSM and "tilting" to SCV by also holding a SCV fund. This is because TSM effectively cancels out all or most of that exposure. An investor who desires to have a meaningful exposure to beta, small, and value can only do that by allocating a larger percentage to SCV, perhaps all of his equity allocation. As I think you've stated elsewhere, a 30% allocation to SCV could give the investor a similar expected return to about 50% TSM with exposure to all 3 risk factors; as well it has less left tail risk because of the smaller percentage held in equities. Is this correct?
We don't know where we are, or where we're going -- but we're making good time.
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Let me get this straight about SCV & factor diversificat

Post by larryswedroe »

DBR
I think I have explained all of those issues over the years, just not in one post.
Also I don't care how you want to define risk adjusted, whether SD or drawdown, low beta and high tilt has won easily over the long term.

Browser
Basically right. Adding this, premiums are time varying not static. So you have to evaluate the expected returns based on today's valuations, not just looking at history. Good example is beta. Don't know anyone who expects future beta of the historical 8.2. And of course bond yields well below historical averages. So what mix you need to get same expected return as TSM changes over time and needs to be accounted for. At any rate no matter what portfolio you choose today you need more factor exposure because yields and beta premiums are way down. That's the bad news but fact/reality.

Larry
dbr
Posts: 46137
Joined: Sun Mar 04, 2007 8:50 am

Re: Let me get this straight about SCV & factor diversificat

Post by dbr »

larryswedroe wrote:DBR
I think I have explained all of those issues over the years, just not in one post.
Also I don't care how you want to define risk adjusted, whether SD or drawdown, low beta and high tilt has won easily over the long term.

Larry
Yes, I agree with you on both points. On the whole you are one source of getting all this straight and that is appreciated.
User avatar
Rick Ferri
Posts: 9703
Joined: Mon Feb 26, 2007 10:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

Re: Let me get this straight about SCV & factor diversificat

Post by Rick Ferri »

Factor investing is becoming quite popular. I haven't seen actual data, but my gut tells me that "smart beta" and "fundamental indexing" concepts are certainly being applied in more portfolios and in smaller accounts. Given these return premiums must come from somewhere, someone has to be taking the other side of the trade. That would be uninformed traders - the alpha seekers.

Here is my dilemma: where are all the new uniformed traders going to come from who are required to provide premiums to a larger set of factor investors? Factor investors believe they have found a smarter beta, thus there must be many more dumb traders to supply these opportunities. Where are they?

II have not seen any evidence that more dumb alpha chasers exist. Thus, it is reasonable to believe that factors may not work as well in the future as the risk premiums must be spread among many more factor investors. Perhaps these premiums will be spread so thin that the cost of factor investing outweighs the benefit in the future.

Just a thought.

Rick Ferri
The Education of an Index Investor: born in darkness, finds indexing enlightenment, overcomplicates everything, embraces simplicity.
YDNAL
Posts: 13774
Joined: Tue Apr 10, 2007 4:04 pm
Location: Biscayne Bay

Re: Let me get this straight about SCV & factor diversificat

Post by YDNAL »

larryswedroe » Fri Sep 05, 2014 9:01 am wrote: To me this is quite simple
1) Since we know that exposure to factors explains almost all returns, then you should construct a portfolio that gives you the exposure to the factors you want in the amounts you want.
2) Using asset classes is "primitive" way of doing it because you can have say three small value funds that are of the same asset class thus many would think are the same, but have very different loadings on factors.
3) And since factors while explaining the vast majority of differences in returns between diversified portfolios have very low correlations it seems logical to seek to diversify across factors---assuming you want exposure to them.

There is nothing wrong with TSM if it gives you the exposure to the factors you want. And of course it's low cost and tax efficient. But you must recognize that while you do own small and value stocks you have no exposure to these factors because the large and growth stocks you own give you exactly offsetting negative exposure to those factors.

So it depends on how you want to look at diversification. (my emphasis)
NO, unless one is hung-up in misusing words. If I were to write the emphasized sentence, I would write "look at exposure to stocks." Of course, you can write (and think) as you wish.

To "diversify" the coins in our pocket, one chooses to have three 25¢ coins, a couple of 10¢ coins, and one 5¢ coin - ALL very different coins, ALL very differently valued at my local grocery store. We can call that TSM for all practical purposes. :)

Code: Select all

Morningstar Valuation:
Value - Core - Growth
  24     24      25    Large
  06     06      06    Mid
  03     03      03    Small
The person choosing the have only 1¢ coins (that would be SCV) in his/her pocket is NOT "diversifying" the coins.
Landy | Be yourself, everyone else is already taken -- Oscar Wilde
User avatar
Clearly_Irrational
Posts: 3087
Joined: Thu Oct 13, 2011 3:43 pm

Re: Let me get this straight about SCV & factor diversificat

Post by Clearly_Irrational »

YDNAL wrote:To "diversify" the coins in our pocket, one chooses to have three 25¢ coins, a couple of 10¢ coins, and one 5¢ coin - ALL very different coins, ALL very differently valued at my local grocery store. We can call that TSM for all practical purposes. :)
The person choosing the have only 1¢ coins (that would be SCV) in his/her pocket is NOT "diversifying" the coins.
Whether you believe in a multi-factor model or not, that just seems like a poor analogy. How do coins possibly relate to future returns?

In a single factor model like CAPM beta explains around 70% of the actual returns.

In a three factor model like Fama-French market, small & value explain around 90% of the actual returns.

In a four factor model like Carhart market, small, value & momentum explain around 91% of the actual returns. (The improvement is actually better than the r^2 value alone suggests)

Source http://www.uv.es/qf/06006.pdf (Lazano 14)

Through extensive testing we know for sure that the multi-factor models are descriptive of past results. Since the original Fama-French papers came out in the early 1990s we now have over 20 years of out of sample testing data that has confirmed their original work so it's quite likely that they're predictive as well.

Given a workable predictive model it behooves us to take heed of the additional factors suggested. Does this mean we would automatically choose a non-TSM portfolio? The answer is no, however to quote Wikipedia "For a nontrivial multi-objective optimization problem, there does not exist a single solution that simultaneously optimizes each objective." There will be a wide selection of portfolios that solve the risk/return preference of any individual investor. For example, one could accept higher tracking error in return for greater return per unit of volatility by emphasizing factors other than market. That choice is by no means mandatory and one could quite legitimately accept the validity of the multi-factor model and still choose a portfolio with no additional loading on factors other than market.
dbr
Posts: 46137
Joined: Sun Mar 04, 2007 8:50 am

Re: Let me get this straight about SCV & factor diversificat

Post by dbr »

Clearly_Irrational wrote: Given a workable predictive model it behooves us to take heed of the additional factors suggested. Does this mean we would automatically choose a non-TSM portfolio? The answer is no, however to quote Wikipedia "For a nontrivial multi-objective optimization problem, there does not exist a single solution that simultaneously optimizes each objective." There will be a wide selection of portfolios that solve the risk/return preference of any individual investor. For example, one could accept higher tracking error in return for greater return per unit of volatility by emphasizing factors other than market. That choice is by no means mandatory and one could quite legitimately accept the validity of the multi-factor model and still choose a portfolio with no additional loading on factors other than market.
That is right. Validity of a multi-factor model for expected return does not suggest any recommendation in particular for how someone might invest. Now Larry Swedroe for one has said more than that, to wit that not just return, but risk adjusted return, can be improved, or that with appropriate allocation to bonds low tail risk can be reduced. I put those things in the category of powerful statements, if they are true, because this starts to incorporate solutions to risk/return that would actually be interesting to investors. My stress is on the fact that you have to make and support those stronger claims before it gets interesting. The FF model does not do that, and I am not sure either Fama or French have gone the extra distance with the analysis, though they may have somewhere.
countmein
Posts: 653
Joined: Fri Dec 06, 2013 8:10 pm

Re: Let me get this straight about SCV & factor diversificat

Post by countmein »

Blue wrote:
countmein wrote:. I happen to believe with more certainty that small value will get absolutely obliterated in deflationary times more than I believe that it will outperform by any wide margin in 'normal times'.
Curious why you believe this?
Institutional and hedge fund ownership of small and micro has skyrocketed (see http://papers.ssrn.com/sol3/papers.cfm? ... id=2416900) in a way that appears more systematic than temporary. Not surprisingly, valuations of both the value and growth sides are extended. Second, we know that risk and return are not related (e.g. SG black hole, Haugen, Falkenstein, etc), so it's not like it's mandatory for small/micro value to do well just because it's risky.
freddie
Posts: 920
Joined: Sat Feb 08, 2014 10:06 pm

Re: Let me get this straight about SCV & factor diversificat

Post by freddie »

YDNAL wrote:
larryswedroe » Fri Sep 05, 2014 9:01 am wrote: To me this is quite simple
1) Since we know that exposure to factors explains almost all returns, then you should construct a portfolio that gives you the exposure to the factors you want in the amounts you want.
2) Using asset classes is "primitive" way of doing it because you can have say three small value funds that are of the same asset class thus many would think are the same, but have very different loadings on factors.
3) And since factors while explaining the vast majority of differences in returns between diversified portfolios have very low correlations it seems logical to seek to diversify across factors---assuming you want exposure to them.

There is nothing wrong with TSM if it gives you the exposure to the factors you want. And of course it's low cost and tax efficient. But you must recognize that while you do own small and value stocks you have no exposure to these factors because the large and growth stocks you own give you exactly offsetting negative exposure to those factors.

So it depends on how you want to look at diversification. (my emphasis)
NO, unless one is hung-up in misusing words. If I were to write the emphasized sentence, I would write "look at exposure to stocks." Of course, you can write (and think) as you wish.

To "diversify" the coins in our pocket, one chooses to have three 25¢ coins, a couple of 10¢ coins, and one 5¢ coin - ALL very different coins, ALL very differently valued at my local grocery store. We can call that TSM for all practical purposes. :)

Code: Select all

Morningstar Valuation:
Value - Core - Growth
  24     24      25    Large
  06     06      06    Mid
  03     03      03    Small
The person choosing the have only 1¢ coins (that would be SCV) in his/her pocket is NOT "diversifying" the coins.
Why buy TSM when you could have some portfolio that gives you 11s in all the boxes which would give you much better Morningstar valuation box diversifications?:) The argument for VBR instead of TSM is that 1.02-.62-.41 is more diversified than 1-0-.03. You can decide if you believe that or not.
dbr
Posts: 46137
Joined: Sun Mar 04, 2007 8:50 am

Re: Let me get this straight about SCV & factor diversificat

Post by dbr »

freddie wrote:
Why buy TSM when you could have some portfolio that gives you 11s in all the boxes which would give you much better Morningstar valuation box diversifications?:) The argument for VBR instead of TSM is that 1.02-.62-.41 is more diversified than 1-0-.03. You can decide if you believe that or not.
Defining diversified as 11% in each of the nine boxes or defining diversified as having a factor loading of 1,1,1 means nothing if there is not also an analysis of what result is produced that is actually of interest to some investor or another. Larry Swedroe actually mentioned two for holding significant positive loadings on small and value factors. I am not sure I have heard a coherent suggestion for the benefit of holding an equal distribution across the M* style box other than the factor loading on small size would be significant as a result. Note that the definitions of size used in the M* box are arbitrary and hence the uneven division of TSM on M* sizes is arbitrary. The definitions of value are set up so that TSM is 1/3 in each. The definitions of FF factor loadings are relative to TSM, so TSM by definition is 1,0,0. That doesn't mean anything other than that it is a definition. Certainly no one should imagine that TSM does not hold small cap stocks or value stocks or small cap value stocks.
lee1026
Posts: 396
Joined: Sat Jun 21, 2014 7:47 pm

Re: Let me get this straight about SCV & factor diversificat

Post by lee1026 »

I'm not debating with you rather the premium is real or not...I'm just simply stating that the 3 funds you described (or that I described) do not have zero correlation with each other. Also, you do realize that a typical small fund has b3~=1 and bs~=0.5 or so and a typical value fund has b3~=1 and bv~=0.5 or so, right?
So you short the "pure value" fund in your "pure small" fund until there is no longer a correlation between each other. Similar things can be done with shorting the TSM fund to remove that correlation.
Through extensive testing we know for sure that the multi-factor models are descriptive of past results. Since the original Fama-French papers came out in the early 1990s we now have over 20 years of out of sample testing data that has confirmed their original work so it's quite likely that they're predictive as well.
Out of sample data have been pretty disappointing since the early 90s. The vanguard value funds have done poorer than their growth counterparts. Some value funds have done better, but an investor also have to ask himself if he have the skills to pick the correct value fund. The vanguard small funds have been roughly even with the large ones in risk adjusted terms.
User avatar
Clearly_Irrational
Posts: 3087
Joined: Thu Oct 13, 2011 3:43 pm

Re: Let me get this straight about SCV & factor diversificat

Post by Clearly_Irrational »

lee1026 wrote:Out of sample data have been pretty disappointing since the early 90s. The vanguard value funds have done poorer than their growth counterparts. Some value funds have done better, but an investor also have to ask himself if he have the skills to pick the correct value fund. The vanguard small funds have been roughly even with the large ones in risk adjusted terms.
What I was trying to convey was that out of sample data has supported the theory, not that small cap value stocks have been the bees knees during that period. The data very strongly supports the multi-factor models over the original CAPM model and it supports that they are predictive and not just descriptive. That said, the variance in factor returns can be large and have extended duration while still conforming to the model.
golfvestor
Posts: 80
Joined: Fri Aug 22, 2014 11:50 pm

Re: Let me get this straight about SCV & factor diversificat

Post by golfvestor »

larryswedroe wrote: clearly, historically a portfolio that has high exposure to factors and then used high quality bonds has produced superior risk-adjusted returns as well as cutting tail risks. That's fact, not opinion.
Larry
You can maybe make that claim for the US, but you can't make it internationally. When Fama did his international factor paper, his start point was Nov 1989. 25 years is a small time period in comparison to how long many countries have had stock markets. What they did for most of their history is unknown and probably unknowable, as accurate p/b and p/e data hasn't been kept except in the very recent past for many countries.
lee1026
Posts: 396
Joined: Sat Jun 21, 2014 7:47 pm

Re: Let me get this straight about SCV & factor diversificat

Post by lee1026 »

What I was trying to convey was that out of sample data has supported the theory, not that small cap value stocks have been the bees knees during that period. The data very strongly supports the multi-factor models over the original CAPM model and it supports that they are predictive and not just descriptive. That said, the variance in factor returns can be large and have extended duration while still conforming to the model.
There is two issues with that. One is that if the factors have the out of sample premium of 0, then there is no good reason to invest in it. Two, according to the same academics, small and value have done well since the early 90s. But the funds that purport to invest in those factors have not done well.
User avatar
Clearly_Irrational
Posts: 3087
Joined: Thu Oct 13, 2011 3:43 pm

Re: Let me get this straight about SCV & factor diversificat

Post by Clearly_Irrational »

lee1026 wrote:There is two issues with that. One is that if the factors have the out of sample premium of 0, then there is no good reason to invest in it. Two, according to the same academics, small and value have done well since the early 90s. But the funds that purport to invest in those factors have not done well.

Code: Select all

Ticker 	mkt 	  smb 	 hml  	 mom 	   (α) 	  R2
VISVX 	0.96 	 0.59 	0.61 	-0.10 	-0.01% 	0.94
VISGX 	1.05 	 0.70 	0.11 	 0.06 	-0.02% 	0.93
VTSMX 	1.00 	-0.00 	0.02 	-0.01 	-0.00% 	1.00

Code: Select all

# 	Initial      Final 	 CAGR 	StdDev 	Best Yr   Worst Yr 	Max. DD  Sharpe  Sortino US Corr Intl Corr
1 	$10,000 	$39,042 	8.89% 	19.58% 	37.19% 	-32.05% 	-36.85% 	0.43 	0.49 	0.71 	0.63
2 	$10,000 	$40,386 	9.12% 	21.63% 	42.88% 	-40.00% 	-40.00% 	0.42 	0.50 	0.91 	0.85
3 	$10,000 	$27,218 	6.46% 	19.93% 	33.35% 	-37.04% 	-37.07% 	0.31 	0.25 	1.00 	0.91

Code: Select all

CAGR/StdDev (as a quick risk/return comparison)
VISVX 0.4540
VISGX 0.4216
VTSMX 0.3241
Everything looks on track to me. What's the perceived problem? Data is from http://www.portfoliovisualizer.com 1998 to 2013
lee1026
Posts: 396
Joined: Sat Jun 21, 2014 7:47 pm

Re: Let me get this straight about SCV & factor diversificat

Post by lee1026 »

I was basing it on the ETFs.

VTV (Vanguard value) have had a mean return of 8.08 with std dev of 15.17 in the last 10 years. (It is new enough that 15 year returns weren't available.) That compares unfavorably to the S&P 500 at 8.38 return and 14.69 std dev.

VB (Vanguard small) have a sharpe ratio of 0.55, barely edging out the S&P at 0.52. This is also for the last 10 years because the fund isn't all that old.

All data is from morningstar.
Post Reply