Bogleheads

[Robert T]

.
Has the value premium been larger and more consistent than the small cap premium?

There have frequently been comparisons between the magnitude and persistence of the value and small cap premiums with the common conclusion that the value premium is larger and more consistent.

However the value premium is commonly calculated as the returns of the top 30% of stocks by book-to-market minus the bottom 30% while the size premium is calculated as the returns on the smallest half minus largest half of stocks (as used by Fama-French). Would the above conclusion change if an ‘apples-to-apples’ comparison is used i.e. if the small cap premium was calculated as 30% smallest minus 30% largest?

The table below presents the results using data from 1927-2006. If the 30:30 approach is used to calculate the small cap premium then:

- The value and small cap premiums were more similar in size (5.23% and 5.88% average premiums; 4.27 and 3.85 annualized premiums respectively)

- The value premium was still more consistent. Of all rolling 10 year periods since 1927 there were only 4 with a negative average value premium while there were 15 with a negative average small cap premium (on annualized basis the respective numbers are 7 and 18.).

- The correlation coefficient of the small cap premium with the equity and value premium was higher using the 30:30 approach than the 50:50 approach

- Another result I find interesting is that on an annualized basis the value premium has been more robust than the equity premium.

I am still comfortable overweighting value relative to small as the value premium has historically been more consistent, has had a lower correlation with the overall market, and has been larger on an annualized basis than the size premium. Obviously no guarantees that this will continue to be the case.

[Trev H]

Robert T...

Thanks for the details on value vs small.

Question for you on Value Options Domestic and International...

EFV for ILV (AMC 44,122) ER 0.40
DLS for ISV (AMC 1,493) ER 0.58

The two above seem to be the clear choices on the International Side.

Looking at US LV choices..

VTV - Vanguard LV (AMC 54,377) ER 0.11
IWD - I Shares R1000 Value (AMC 48,146) ER 0.20
RPV - Rydex S&P 500 Pure Value (AMC 12,590) ER 0.35

US SV Choices

VBR Vanguard SV (AMC 1,562) ER 0.12
IWN I Shares R2000 Value (AMC 1,093) ER 0.25
RZV Rydex S&P 600 Pure Value (AMC 597) ER 0.35

The I Shares ETF's are a bit more costly than VG ETF's for LV and SV.
Then the Deep Value funds are even more costly (but still quite a bit different than the standard LV SV).

I am considering a 4 way split on US Value.

VTV, RPV for US LV
VBR, RZV for US SV

Based on the AMC of each they seem to be covering quite different areas of the Value Market. Using the VG ETF's would also lower my overall cost of LV SV in my Portfolio.

I understand from a previous post that you use mostly the Russell Indexes for Value.

IWN cost a bit more than twice as much as VBR and the AMC is quite a bit closer to RZV than VBR is... which makes me favor the combo of VBR and RZV for US Small Value.

IWD is near twice as expensive as VTV.... but both IWD and VTV cover quite a bit different area than RPV does.... so based on coverage... I could go either way on the LV Combinations... VTV & RPV or IWD and RPV.

I have a personal preference for Equally Weighting slices in my portfolio...
I will be holding 1 part US Large Cap Blend + 2 parts US Large Value with the LV split between (one of the two combinations above).

May do the same on Small...

1 part Small/Micro Blend, 2 parts SV (VBR & RZV).

I understand (again from a previous post) that you use BRSIX for US Micro Cap exposure.

Have you considered IWC which has a AMC of 379 and ER of 0.60 ?

You may be using BRSIX instead if you are holding it in a taxable account.

My investments will be in a IRA so no tax issues.

I just noticed that BRSIX cost a bit more than IWC 0.69 vs 0.60 and BRSIX leans towards SG a bit... where IWC is SB.

Anyway...

Would like to hear your comments on using the VG ETF's for Standard LV & SV vs the I Shares ETF's.... and what you think of BRSIX vs IWC for US Microcap Exposure.

I have not moved to Wells Fargo yet... but I do expect to soon and have been looking over the ETF options and trying to make some final decisions at this point.

Thanks

Trev H

[larryswedroe]

This is point I have made many times.

But don't forget that it is much easier to load on size than on value (perhaps because of the very issue of difference in definition). DFA micro for example loads over 100% on small but value is well below 100%.

Also with small can add value with screens and block trading--not so much with value.

In end it is the net returns you get.

One last point, there is some behavioral errors on small stocks---lottery effect, leading to overvaluation (small growth anyway), but behavioral impact on value leads to undervaluation. Or that free stop at the dessert tray. At least historically, though perhaps no more now that most of world knows about this

[ramsfan]

[alvinsch]

[Robert T]

.
Trev H,

Let me have a try at your questions…

As indicated in an earlier post – I find the best way to try answer is just to share what I tried to do on the same/similar questions (right or wrong). This sometimes gives less than full answers but at least gives what I know or did…

First select portfolio factor loading targets: What I find has helped in security selection is first to set portfolio factor loading targets i.e. how much of the value and size premiums (if they exist) do I want to try capture. My choice was based primarily on need to take risk (but also on willingness and ability to take risk), which ended up being 40% of the value premium and 20% of the size premium as defined by Fama-French.

Second select securities to achieve portfolio factor loading targets: With these targets set, the criteria used for security selection (at least for me) is primarily how to achieve these factor loading targets at minimum cost (considerations on diversification, style purity, and fund stewardship are also important). I also prefer diversification across asset classes within the parameters of my loading targets – this should not change expected returns but should reduce asset class tracking error regret. In an earlier post Larry summarized this well:

“--it is the loading factors that matter, not exposure to each asset class. However owning each often has benefit of keeping investors disciplined and practiced at rebalancing---a benefit IMO.”

Intl. value: iShares EAFE Value seem to be a reasonable option (which I use). I have not looked at DLS in any great detail as at this point I don’t need extra value exposure to achieve my portfolio value loading targets (and have not found the historical back-tests on DLS to analyze). I currently use Vanguard Intl. Explorer but may consider an intl. small value fund in the future to get a more even spilt in my domestic and international factor exposure – not to change my overall portfolio targets. As I understand DLS is the only intl. small cap value fund available to do-it-yourself investors so there seems little alternative if you want intl. small value exposure.

US Large value: IMO a large value fund should have a close to zero size loading and high value loading. Unfortunately many (most) large value funds (including the Vanguard Value Index) have negative size loadings and so subtracts from an overall portfolio objective of tilting away from the market to value and small cap companies. So my choice for ‘large value’ was the iShares S&P400v ETF – it has a small but positive size loading and a relatively large value loading. I am not sure what the size loadings are for IWD and RPV but would expect them to be negative (but could be wrong).

US Small value: My first choice was iSharesS&P600v when I setup my portfolio. It was formed on price to book, had low expenses, was relatively tax efficient, and run by BGI so reasonable fund stewardship). However in Dec 2005 its benchmark construction changed from a sole price-to-book to a multifactor criteria (as with the iShares 400v). This reduced the value loadings of the funds and hence the value loading of my portfolio. To get back to my factor loading targets I added RZV (in tax advantaged account).

I tried to review some small value ETFs in an earlier post on M* 57743. My understanding is that the recent change in benchmark on the Vanguard Small Value fund to the MSCI small value index has raised its value loading. If this is all that’s needed to meet a portfolio value loading target in an IRA account then it’s a good choice IMO.

US Microcap: When I set-up my portfolio the Bridgeway fund had one of the lowest weighted-average market capitalization of any passively managed fund. It is also managed to minimize/lower capital gains distributions and hence taxes and transaction costs. The approach however lowers fund turnover relative to its quarterly rebalanced benchmark leading to relatively high tracking error. I continue to hold the fund. My view is that a micro-cap fund is the most efficient way (takes up the least space) to capture the small-cap premium if it continues to exist (although I think the Vanguard Tax-Managed Small Cap Fund is also excellent) [see M* 52926 #51 for an earlier analysis of small versus micro). I have not considered IWC. Following the primary security selection criteria above, I don’t think it will reduce the cost of achieving my portfolio factor loading targets (give there will be some cost to the change). It will be interesting to see how IWC performs.

So my view is that only one value fund is necessary to achieve a portfolio value loading target – any additional value funds will not improve expected return if targets are maintained but may address asset class tracking error regret (the reason I have ‘large’ and small value). So it may not be necessary to add two large value and small value funds as you suggest. As above the primary reason I currently own two small value funds was to get back to my value loading target following the change in the iShares benchmark.

Just a modest effort. Not suggesting this approach is right for everyone but it has been tremendously helpful to me.

_________________________

Larry:

I agree its easier to load on small than value, but if a portfolio value loading target is relatively small then it can be achieved by adding existing value funds – so don’t have to compensate by adding more exposure to micro-cap funds to get target expected portfolio return. I agree on the lottery effect of small-growth stocks.

_________________________

Al:

A plot of the rolling 10 year periods of the annualized value premium shows it to be remarkably consistent (at least to me) with relatively small deviations around about 4 percent particularly from 1953-93. The 10 year periods that included the great depression years showed negative value premiums as did the period to the later 1990s. A trend line plot including all periods showed a slightly positive trend (perhaps not significant) although this was influence by the negative periods at the start of the series in the late 1920-early 1930s.

A similar plot of the size premium 10 year periods shows a much larger variation and what appears to be less robust (consistent) premium. There were large premium peaks in the 10 years to 1940, 1968 and 1983, but large negative 10 year periods to 1956 and 1998. A trend line plot shows a downward slope perhaps influenced by 14 consecutive negative rolling 10 annualized size premium periods from 1989 to 2002


Robert
.

[Black Knights]

Where does one find SmB and HmL for international funds? Without these, how does one determine their portfolio's loading points?
_________________
Jed

[Robert T]

.
Black Knights,

For Intl HmL go to:

1. http://mba.tuck.dartmouth.edu/....brary.html

2. Under section on International Research Returns Data (Downloadable Files) open Index Portfolios formed on B/M, E/P, CE/P, and D/P

3. Open Ind_all

4. Subtract Low BE/ME from High BE/ME to get High BE/ME minus Low BE/ME or HmL.

This is the Intl HmL I used to estimate value loadings on Non-US developed market value funds and use it to construct this part of my portfolio benchmark.

For Intl SmB were on our own:

One approach is to go to the MSCI website and just use EAFE small cap index minus EAFE.

Robert
.

[SmallHi]

Robert,

Interestingly enough, we seem to often be doing the same or similar research only weeks apart of eachother (you are much more apt to post on it than I am).

An interesting quirk I added to this research is to look at the historical magnitude and consistency of 2 "homemade risk dimensions", namely:

TSM minus 5YR T-Notes,

AND

SV minus TSM

What I found would probably surprise most investors. That is, SV has beat TSM with about the same magnitude and consistency over 1, 3, 5, 10, 15, and 20 year rolling periods as TSM has beaten 5YR T-Notes (one often uses 5YR T-Notes to illustrate the stable bond portion of their portfolio).

What I came away thinking is, there really are a number of ways to strucutre an investment portfolio to the various risk factors that get you to a very similar end result.
For example, to use your personal portfolio for a minute:

(0.75, 0.2, 0.4, 0.5, and 0.0)
for beta, size, price, term, and default --

could very easily be done with:

0.6, 0.5, 0.7, 0.5 (or 0.2 ), and 0.0

The latter portfolio consists of equal parts LV and SV, but a heathier dose of fixed income (and potentially less term risk using 1 or 2 YR maturities).

Ultimately, I don't think one is better than another, but best for a particular person dependent on their tolerance for various risks (volatility, tracking, human capital...)

smallhigh

[Robert T]

.
SmallHi,

Portfolio structure: I fully agree that there are many different ways to structure a portfolio for the same expected returns. As Larry indicated earlier– his 25:75 small value:fixed income portfolio has about the same expected return as a 60:40 TSM:fixed income. While the expected returns are similar, what seems to differ is the expected dispersion of return (ie: fatter or thinner tails).

Re: your example: I find that a 0.7 portfolio value loading is difficult to achieve outside of DFA funds (while it maybe possible with the Rydex funds for the US allocation, I would not be comfortable with all my value exposure in them [personal preference]; for non-US developed the value loads are usually smaller than what’s available on the US side and for EM the options are even scarcer).

Personalization:As you indicate – personalization matters. As Ellis puts it in “Winning the Loser’s Game”: “To be a truly successful lifetime investors, the first and central challenge is to ‘know thyself” – and I would add “and situation”. This can impact almost all allocation decisions and certainly has impacted mine. I am comfortable with my resulting risk allocations.

Non-linearity: Re: an earlier post of yours on linearity of risk factors – which I missed (can’t keep up: ). I think there are non-linearities in the risk factors to varying degrees. Less so for value, somewhat for small, and more so for term (quadratic variables are often added to models of the term structure). However I don’t think the additional complexity of trying to model non-linearity is worthwhile, and the linear model already explains about 95 percent of return variability. Use of microcap over smallcap stocks, and intermediate term over long-term treasuries is the only way I currently (although only partially) reflect non-linearity in my allocation. Term choice was also influenced by other factors (as dicussed earlier).

Default risk (somewhat linked to non-linearity): The attached paper (which may also be of interest for Trev H) was useful in my initial allocation decisions and reflected in IPS (it subsequently came out in Journal of Finance - both versions were useful so they are both linked).

Default Risk in Equity Returns [working paper]
Default Risk in Equity Returns [published version]

Two important messages from the papers (at least to me).

- There is some default risk in equity returns (particularly in micro-cap and value stocks) which seems consistent with the cost of capital story and with the significant correlation of SmB with bond default risk (as in the FF common risks paper). So IMO there is no need to take default risk in fixed income – its to some extent already reflected in equity returns and if included, fixed income would have higher correlation with equities which goes against my role of fixed income.

- Default risk has most significant effect on returns of CRSP9-10 and 10. Hence the use of a microcap fund with CRSP10 benchmark.

Robert
.

[SmallHi]

Robert --

Interesting thoughts. I agree with all of them. I would like to spend some time "checking up" on the linearity of value returns when I get a chance to check the French 5X5 data.

I have seen some measures in FF papers that break size and value down much finer than the 3X3, and that value linearity persists.

But, when viewing the simulated S&P Pure Value index returns, we run into some issues. For example, despite the extreme value loading of S&P 600 Pure Value (1.0 vs. 0.8 for FF SV), it has a severely negative alpha (almost statistically significant at -1.7).

Also, (while I would put much less weight on this) the FF SV Index compounded at almost 4% more per year than the S&P 600 Pure Value index since 1995....which is just another way of demonstrating the negative model alpha.

All this, at least over this time period, may indicate there is not complete linearity in HmL. This is somewhat important in my mind....as it would render these Pure Style indexes obsolete for all but the extreme value tilter (as the moderate tilter would most likely prefer to get their size/value exposure in a more diversified Russell/MSCI/S&P NON Pure Style Fund or ETF product)

smallhigh

[Robert T]

.
SmallHi,

Few more thoughts.

1. Negative alpha on value funds: The negative alpha is not unique to the S&P Pure Value index but seems common across most value indexes (see table below). Arnott highlighted this in his ‘Fundamental Indexation’ paper “…most value indexes earn an estimated Fama–French alpha of –1.5 percent or worse” (presumably per year). The negative alphas are difficult to fully interpret. The easy route is just to say they are statistically not significantly different from zero (which is often the case, except for the alpha on the Russell 2000 value index) so no need to worry. Bernstein suggests its just a flaw in the model “the alphas of these indexes, which are statistically significant, represent a flaw in the model, which does an otherwise excellent, but obviously not perfect, job of predicting an extraordinarily complex system using just three variables. The Three-Factor Model is, as Fama and French have repeatedly emphasized, only a model — it is not reality.” Arnott attributes the negative alpha to poor construction of the indexes. Perhaps there is some truth to both these views. For example the significant negative alpha on the Russell 2000 value index maybe capturing the effects of the reconstitution arbitrage which seems widely acknowledged (see for example Swensen’s Unconventional Success pg. 251-257 on the Russell index construction). IMO other indexes being tracked by mutual funds may also not be immune to this effect.

2. Linearity in value and small cap premiums: The size premium seems largest in the smallest decile by market cap, while the value premium seems largest in the highest 30% of BtM (see table below) – so slightly more linearity in the value premium. If we assume that the pattern will continue then it seems to suggest: (i) Micro-cap funds which target the CRSP10 may be the most efficient way to capture the size premium; and (ii) value funds which target the highest third BtM may be the best way to capture the value premium.

This conclusion is similar to that in the Default Risk in Equity Returns paper in the earlier post. I am only aware of one fund that targets the CRSP10 – the Bridgeway Ultra-Small Market Fund (although it seems fairly difficult to closely track the index). Most indexes (S&P, Russell, MSCI) use close to a 50:50 split to separate ‘growth’ and ‘value’ which will still capture some of the value premium. My understanding is that DFA’s value funds target stocks in the highest 30% BtM and the S&P Pure value indexes target the highest 33% by ‘value score’ (using multiple criteria and then sort by score).

[sterjs]

Robert,

Did you compute the factor loadings, turnover, etc on an annualized basis or just use the current factor loadings?

I would expect DFA to be the least tax efficient choice because it lacks an ETF share class. I have concerns about VBR on that front as well--the mutual fund shares dwarf the ETF shares.

[Robert T]

.
sterjs,

Factor loadings were computed using monthly data from July 1995 to April 2007.

Turnover was calculated as a three year average based on data availability.

- For VBR, the 2004-06 average annual turnover for the Vanguard Small Value Index Fund was used [the annual turnover was 30% for 2004, 28% for 2005, and 25% in 2006 to give an average of 28%]. Data prior to 2003 reflect more the turnover of the previous index tracked by the Vanguard fund and not the MSCI index.

- For IJS and RZV, the 2002-04 annual average turnover for the simulated S&P series was used. These were provide for both the Citigroup and pure value indexes in the earlier S&P paper on “Unveiling the Next Generation of Style Indexing”. The annual breakdown was not provided in the paper and this was the only time period average provided.

- For IWN, the 2002-04 annual average turnover for the ETF was used [the annual turnover was 26% for 2002, 45% for 2003 and 16% for 2004 to give an average of 29%]. The same period as used for the Citigroup and pure value series was used for comparative purposes – although the IWN annual turnover is as at March 31 of each year.

- For DFA Small Value, the 2002-04 annual average turnover was used [the annual turnover was 26% for 2002, 35% for 2003, 30% for 2004 to give an average of 30%]. The DFA annual turnover data is as at November 30 of each year.

Dividend yield – was the dividend yield as of Feb. 2007.

% REITS – was the REIT composition as of Feb. 2007.

Management fees – use the current expense ratio. These may change in the future. As per the prospectus of RZV – “The Fund has adopted a Distribution (12b-1) Plan pursuant to which the Fund may be subject to an annual 12b-1 fee of up to 0.25%. However, no such fee is currently charged to the Fund and no such fees will be charged prior to March 1, 2008.”

Robert
.

[sterjs]

[DaveTH]

[SmallHi]

Robert,

Thanks for the continued efforts. I am certainly bookmarking these posts.

Quick question, are your SV Index returns reduced by current expense ratios? (I don't assume they are, and not that it matters, just wanted to check)

Also, I should probably clarify my comments above about the FF & S&P 600 Pure Value return differential of almost 4% since 1995. I was using the Fama French Small High series, not the Small Hi (xUtilities) series that DFSVX is benchmarked to (an error on my part). Over this period, the performance difference was significant (almost 3% a year to the former), and much higher than the previous 7 decades.

From 7/95 through 4/07 I have DFSVX compounding at a hair over 16.6% annually, but its no big deal. My regression estimates are dead on with yours.

(Interestingly, over this period, DFSVX net of fees outperformed FF SV (xutilities) by about 0.4% a year! Thats some transaction skill!

Finally, I find it interesting that over this period, FF US Vector had annualized returns higher that every retail SV Index except the 600 Pure Value (despite lower size/value exposure) -- 14.9% a year, yet with a correlation to the market of almost 0.9 vs. 0.7ish for the SV funds.

OK, enough of my S/V ramblings...

edit: i went back, and rechecked my #s, and I think Robert's SV Index returns are from 1996-2006 and not the 7/95-4/07 that I assumed -- this throws off my quoted #s just a bit...

thanks!

[mptfan]

SmallHi, I have done quite a bit of study regarding the French/Fama three factor model lately, and I am pretty convinced that a small / value tilted portfolio which includes U.S. and international equities is the way to go.

I am considering using the following allocation:

10% U.S. LCB
10% U.S. LCV
10% U.S. SCB
10% U.S. SCV
10% U.S. Microcap
10% Int'l LCV
10% Int'l SCB
10% EM Blend
20% Total Bonds

How do I figure out the size (SmB) and value (HmL) load factors of this portfolio? Is there a source to find the size and value load factors for individual funds that I can reference? Thanks.

P.S. I want to thank you and Robert T and Trev H and DaveTH, and others on this forum who have greatly aided me in my education.

[SmallHi]

MPTfan,

what funds/indexes are you using to track the various asset classes? There is a fair amount of difference from one to the next.....

smallhi

[mptfan]

SmallHi, here are my proposed funds:

10% SNXFX (LCB)
10% VTV (LCV)
10% SWSMX (SCB)
10% VBR (SCV)
10% IWC (Microcap)
10% SWINX (Int'l LCB) (I know I said value earlier)
10% DLS (Int'l SCV)
10% VWO (EM)
20% SWLBX (Total Bond)

As you can tell, most of my funds are at Schwab.

[sterjs]

Is there a reason you hold those Schwab funds? Those ERs are outrageous for cap-weighted index funds.

also, i doubt Fama-French would support total bond marke for bond allocations based... I think they try to keep bonds high credit and short term.

[mptfan]

[psteinx]

Hey, I know this is an old thread, but there's some good stuff above, and I'd like to follow up with more questions.

1) When did Vanguard switch from whatever its previous index target was to MSCI US Small Value for VBR/VISVX?

2) Discuss the merits of the 3 main small value indexes used.

Currently, I own IWN (bought a bit over 2 years ago). But if I understand correctly, the Russell 2000 value is subject to relatively high turnover and the potential for front running, because it has a hard upper and lower boundary, and mechanical, once-a-year recomposition. In contrast, a well designed small value index would hopefully have buffer ranges to reduce turnover, and might perhaps use other techniques as well. Does MSCI Small Value do this? How about S&P 600 V?

3) How will these funds likely compare on tax efficiency? IIUC, ETFs generally have an advantage there, but the Vanguard ETFs act more like open ended funds, because the ETF offering is a small part of the overall Vanguard small-value fund. Is my view correct? Any other thoughts on this?

[hafis50]

It's amazing how different the findings of the data-miners can be:

I recall Dimson/Marsh found a strong long-term value premium in the UK.
[Edit: They also found a strong underperformance of European small caps over the preceding 10-15 years which is supported by the index data published by Stoxx]

This paper finds a small cap premium but no value premium in Europe between 1974 and 2000:

Annaert - Value and Size Effect: Now You See it, Now You don't, University of Gent, Working Paper, 2002

[Michael Weiss]

First, here is MSCI's US Index methodology:

http://www.msci.com/us/MSCI_Mar07_USMethod.pdf

Small premium versus value premium is fascinating topic.

Some may disagree, but IMO the FF method of grouping value/growth stocks is not a good way to select growth stocks. IMO, any study using book to market as a way to select growth stocks will exaggerate the return disparity between the two styles.

[Robert T]

.

[psteinx]

If I understand correctly, the Russell indexes use no buffer zones. Given that the Russell indexes are now being used as the basis for ETFs, is there any indication that Russell will switch and start using buffer zones?

[Michael Weiss]

[psteinx]

Yes, it looks like Russell is adding some buffering to their indexes - see this and Russell's own info here.

I'm still not sure they're using best practices. According to the links, the buffer zones seem rather small, may not apply to the bottom of the index, and there's no indication of any buffering of the style components (i.e. ideally, a value stock that barely ekes its way across the 50th percentile divide would not necessarily be sold off).

[psteinx]

Here's a paper from Russell stating that for 2000-2006, the effect of reconstitution front-running trading and the like hurt Russell 2000 returns by 0.08% per year, on average. That's non-trivial, but it's also not THAT big. They claim the effect is diminishing in recent years.

[Michael Weiss]

[Robert T]

.
A couple more comparisons of the 3 (parent) indexes:

Using small-cap-index ETFs to plug portfolio gaps
MSCI's New US Equity Index Family

And a couple more studies on the 'front running' effect on the Russell 2000 index

Robert

[tc33]

Robert

Are your numbers adjusted for risk (as measured by SD)? From what I understand (and could be wrong), the reason small and value were found to be premiums by Fama & French is because small and value have produced higher risk-adjusted returns over stocks with the same beta (in the old CAPM world).

Without a discussion of risk along with return, I can't see how we're validating that the premium had continued to exist. Yes, small has done better than large, and value has done better than growth, but what about on a risk-adjusted basis?

Again, I could be completely misunderstanding the concept, and would appreciate any education you can provide.

Thoughts? Thanks so much

Tom

[wearethefall]

I have just finished reading David Dreman's book Contrarian Investing: The New Generation. In it he states that much of the size premium prior to 1945 is an illusion. High bid-ask spreads of up to 50% and very low liquidity meant that the data does not accurately reflect the viability of pursuing a small-cap strategy.

Dreman is however, a firm believer in the value premium for behavioural reasons.

[Michael Weiss]

[Kenster1]

But the interesting thing is that based on the M* Tax Cost Ratio for the past 5 years ending 9/30/07, Vanguard's SV (Open-end) Fund has a lower tax-cost ratio than IJS (iShares S&P600 SV) --- 0.38% vs 0.41%.

So I have the M* Tax-cost ratio as (lower the better):

VISVX... 0.38% (Vanguard SV OEF)
IJS........ 0.41% (iShares S&P600 SV)
IWN...... 0.54% (iShares Russell 2000 SV)

Also for Largecap Value, here's the M* 5-yr Tax Cost Ratio:

VIVAX... 0.39 (Vanguard Large Value OEF)
IVE....... 0.69 (iShares S&P500 Value)
IWD...... 0.79 (iShares Russell 1000 Value)

"Tax-adjusted returns and tax cost ratio are estimates of the impact taxes have had on a fund. We assume the highest tax rate in calculating these figures. These returns follow the SEC guidelines for calculating returns before sale of shares. Tax-adjusted returns show a fund’s annualized after tax total return for the three-, five-, and 10-year periods, excluding any capital-gains effects that would result from selling the fund at the end of the period. Fund loads are also subtracted from the figure. To determine this figure, all income and short-term capital gains distributions are taxed at the maximum federal rate at the time of distribution. Long-term capital gains are taxed at a 15% rate. The after tax portion is then assumed to be reinvested in the fund."
_________________
SURGEON GENERAL'S WARNING: Any overconfidence in your ability, willingness and need to take risk may be hazardous to your health.

[Robert T]

.
Tom,

[Michael Weiss]

[Robert T]

.
Kenster,

You raise an important point on who’s data do we use – those from M* or from the fund companies themselves.

Here are the side by side comparisons of data from each of the fund websites and that on M*. I have also included the M* tax cost calculation to all. The main difference is in the after tax returns for ETFs.

Who do we believe? I currently go with the data from the fund companies (as presented in the earlier post). If we compare the 3-yr tax efficiency of VISVX and its ETF share class VBR, M*'s numbers suggest VISVX has superior tax efficiency (tax cost ratio of 0.37 versus 0.58.). If we use Vanguard’s data the tax efficiency is similar (0.38 for VISVX versus 0.40 for VBR) which is what I would expect.

[tc33]

Robert, thanks for the response.

[tc33]

I realized that Yahoo finance has 3-year as well as longer term 5- and 10- year Sharpe ratios in its database, so let's examine the recent relationships between small cap stocks and risk-adjusted returns:

Funds:
Micro Cap: DFSCX - DFA US Micro Cap
Small Cap: DFSTX - DFA US Small Cap Portfolio
Large Cap: DFLCX - DFA S&P 500

[Robert T]

.
Tom,

Apologies, I must have misunderstood your question and I’m still not sure I understand it in relation to the first post but hopefully this lands somewhere close.

On my first post: What I tried to do with the analysis was answer the question I posed on size and consistency of an (apples to apples) comparison of the small cap premium relative to the value premium.

In previous FF analysis they state

- “the average SMB return is only 0.27% per month…HmL produces and average premium of 0.40% per month.” (from Common Risk Factors in the Returns on Stocks and Bonds which used data from 1963-1991)

- “the average HML return is large for the full 7/29-6/97 sample period (0.46 percent per month.)…the average SMB return for 7/29-6/97 is 0.20 percent per month. (from Characteristics, Covariances, and Average Returns:1929-1997, with Davis)

They conclude from the use of HmL and SmB (as I used in the post) that “The value premium in average stock returns is robust” and “The size effect in average returns in smaller”.

I was simply looking at whether this was true if an apples-to-apples comparison of HmL and SmB were made – nothing more (specifically comparing returns of 30% smallest minus 30% largest by market cap rather than the 50%:50% comparisons used by FF. This is more consistent with the HML definition of returns of 30% highest BtM minus 30% lowest BtM). The consistency of these returns over rolling 10-year periods were also presented.

On-Beta: