Performance update - Vanguard, iShares, DFA, RAFI

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
Topic Author
Robert T
Posts: 2803
Joined: Tue Feb 27, 2007 8:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Performance update - Vanguard, iShares, DFA, RAFI

Post by Robert T »

.
FWIW I have just updated some performance numbers on portfolios with similar factor exposure (similar expected return). The portfolios include both ETFs (Vanguard, iShares), Bridgeway, DFA, and (RAFI) Powershares. Few observations:
  • - Portfolios with similar factor loads had similar portfolio returns
    - There are a range options for individual investor to self-implement a global factor targeted portfolio
    - The RAFI (simulated) series perform fairly well, time will tell if Powershares can implement effectively.

Code: Select all

ETF Sample portfolio                DFA Tax-Managed Sample portfolio
----------------------------        -------------------------------- 
23% Vanguard Midcap Value            17% DFA TM US Marketwide Value
 9% iShares S&P600 Value              7% DFA TM US Targeted Value
 5% Bridgeway Ultra Small Co.        13% DFA TM US Small Cap
23% iShares EAFE Value               17% DFA TM Intl Value
 5% iShares EAFE Small               11% DFA Intl Large Cap
10% Vanguard EM                      10% DFA EM  
25% iShares 3-7 yr Treasury          25% DFA Intermediate Government


ESTIMATED FACTOR LOADS
                               Size load   Value load   Term load  
ETF Portfolio 
------------ 
  US equity                       0.36        0.57 
  Non-US developed equity         0.07        0.31 
  EM equity                       0.00        0.00  
  -------------------------------------------------------------- 
  Total equity                    0.21        0.40 
  Fixed income                                             0.50 

DFA Tax-Managed Portfolio 
------------ 
  US equity                       0.40        0.57 
  Non-US developed equity         0.02        0.30 
  EM equity                       0.00        0.00 
  -------------------------------------------------------------- 
  Total equity                    0.21        0.40 
  Fixed income                                             0.60


            ETF         DFA Tax-Managed      Personal
         Portfolio         Portfolio         [Actual]
 
2003        35.8              36.0             36.5
2004        19.4              17.6             18.6
2005        11.7              11.4             12.7 
2006        18.1              18.8             18.3
2007         5.3               7.9              7.0
2008       -27.0             -28.5            -28.7  
2009        27.5              27.2             27.7
------------------------------------------------------
AR          11.2              11.0             11.2              
SD          20.2              20.6             20.8
------------------------------------------------------

Code: Select all

RAFI Sample portfolio               DFA Core/Vector Sample portfolio
---------------------------         --------------------------------

19% P’shares FTSE RAFI US            37% DFA US Vector  
18% P’shares FTSE RAFI US Small      28% DFA Intl Vector
14% P’Shares FTSE RAFI Intl          10% DFA EM Core
14% P’Shares FTSE FARI Intl Small    25% DFA 5yr Global
10% P’Shares FTSE RAFI EM
25% iShares 3-7 yr Treasury


GUESS AT FACTOR LOAD ESTIMATES
  
                              Size load   Value load     Term load  
RAFI Portfolio 
------------ 
  US equity                       0.35        0.35 
  Non-US developed equity         0.35        0.35 
  EM equity                      -0.10        0.35  
  -------------------------------------------------------------- 
  Total equity                    0.30        0.35 
  Fixed income                                             0.50 

DFA Tax-Managed Portfolio 
------------ 
  US equity                       0.40        0.40 
  Non-US developed equity         0.40        0.40 
  EM equity                       0.20        0.20 
  -------------------------------------------------------------- 
  Total equity                    0.37        0.37 
  Fixed income                                             0.20

Simulated returns:

            RAFI          DFA Core/Vector    
 
2003        42.8              41.6             
2004        19.3              19.5             
2005        13.6              12.2             
2006        19.6              19.6             
2007        10.3               6.1             
2008       -27.6             -30.1            
2009        37.5              32.0            
-------------------------------------------
AR          14.3              12.1            
SD          22.9              22.9            
--------------------------------------------

The RAFI data is from Arnott’s Fundamental Index book, supplemented with data from Powershares. 
  • For estimates of US RAFI value and size loads see earlier estimates Arnott (slide 27), which were similar to Bernsteins, the same are assumed for Intl markets.
Robert
Last edited by Robert T on Tue Jan 26, 2010 9:45 pm, edited 1 time in total.
User avatar
Rick Ferri
Posts: 9703
Joined: Mon Feb 26, 2007 10:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

Post by Rick Ferri »

- Portfolios with similar factor loads had similar portfolio returns
- There are a range options for individual investor to self-implement a global factor targeted portfolio
Good analysis and summary, Robert. Thanks.

What this shows is that a portfolio's tilt to value and small cap factors have the greatest impact on risk and return, not the funds used to implement those tilts, i.e. Vanguard or DFA didn't matter. I dismiss RAFI because the returns are based on hypothetical returns from the index provider and not actual funds.

Rick Ferri
caklim00
Posts: 2420
Joined: Mon May 26, 2008 10:09 am

Re: Performance update - Vanguard, iShares, DFA, RAFI

Post by caklim00 »

Robert T wrote:

Code: Select all

ETF Sample portfolio                DFA Tax-Managed Sample portfolio
----------------------------        -------------------------------- 
23% Vanguard Midcap Value            17% DFA TM US Marketwide Value
 9% iShares S&P600 Value              7% DFA TM US Targeted Value
 5% Bridgeway Ultra Small Co.        13% DFA TM US Small Cap
23% iShares EAFE Value               17% DFA TM Intl Value
 5% iShares EAFE Small               11% DFA Intl Large Cap
10% Vanguard EM                      10% DFA EM  
25% iShares 3-7 yr Treasury          25% DFA Intermediate Government


ESTIMATED FACTOR LOADS
                               Size load   Value load   Term load  
ETF Portfolio 
------------ 
  US equity                       0.36        0.57 
  Non-US developed equity         0.07        0.31 
  EM equity                       0.00        0.00  
  -------------------------------------------------------------- 
  Total equity                    0.21        0.40 
  Fixed income                                             0.50 

DFA Tax-Managed Portfolio 
------------ 
  US equity                       0.40        0.57 
  Non-US developed equity         0.02        0.30 
  EM equity                       0.00        0.00 
  -------------------------------------------------------------- 
  Total equity                    0.21        0.40 
  Fixed income                                             0.60


            ETF         DFA Tax-Managed      Personal
         Portfolio         Portfolio         [Actual]
 
2003        35.8              36.0             36.5
2004        19.4              17.6             18.6
2005        11.7              11.4             12.7 
2006        18.1              18.8             18.3
2007         5.3               7.9              7.0
2008       -27.0             -28.5            -28.7  
2009        27.5              27.2             27.7
------------------------------------------------------
AR          11.2              11.0             11.2              
SD          20.2              20.6             20.8
------------------------------------------------------
Congrats on tracking the DFA portfolio so closely. Very impressive. I tilt more on the US side like you, but not nearly to your level as I try to get a bit more small/value in international using DLS/DGS. I seem to remember you being mostly taxable though...
User avatar
RaleighStClaire
Posts: 733
Joined: Mon Jun 18, 2007 2:08 pm
Location: Tilting

Post by RaleighStClaire »

Very nice Robert. I think we all appreciate these comparisons.

Question: How has Schwab been performing on their RAFI funds in comparison to Powershares?
Where's that red one gonna go?
bhzmark
Posts: 172
Joined: Mon Feb 02, 2009 7:48 pm

Post by bhzmark »

Thanks for sharing the info. But wouldn't it be more useful to just compare the equity allocations as 100% equity allocations?

And fixed income assets are a different comparison and analysis?

In any case, thanks for doing the work and sharing it.
Lucio
Posts: 286
Joined: Wed Apr 08, 2009 10:31 pm

Re: Performance update - Vanguard, iShares, DFA, RAFI

Post by Lucio »

Robert T wrote:.
FWIW I have just updated some performance numbers on portfolios with similar factor exposure (similar expected return). The portfolios include both ETFs (Vanguard, iShares), Bridgeway, DFA, and (RAFI) Powershares.
Robert,

How do you estimate the factor loads for the international equity portfolios? Isn't it problematic to use a benchmark series that incorporates multiple markets, currencies and tax regimes?

Lucio
Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Post by Rodc »

bhzmark wrote:Thanks for sharing the info. But wouldn't it be more useful to just compare the equity allocations as 100% equity allocations?

And fixed income assets are a different comparison and analysis?

In any case, thanks for doing the work and sharing it.
My preference is to include FI, as that is what I do in real life. There is an interplay between equity and FI and I like it in the evaluation.

There may be value in looking at things such as stocks in isolation from other factors if one is trying to understand how stock markets work. But as an investor, I prefer to think about things at the portfolio level.
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.
User avatar
Topic Author
Robert T
Posts: 2803
Joined: Tue Feb 27, 2007 8:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Post by Robert T »

.
Rick Ferri wrote:I dismiss RAFI because the returns are based on hypothetical returns from the index provider and not actual funds.
I somewhat agree. Even the series provided in Arnott’s book (the source of much of the data used) are slightly different from the indexes that the Powershares funds track. The DFA vector series are also simulated. Nevertheless I think they are instructive. The RAFI international funds, by their construct, offer the potential for larger country ‘rebalancing’ returns IMO – as in this earlier analysis: http://www.bogleheads.org/forum/viewtop ... t=intl+hml Time will tell, obviously no guarantees.
caklim00 wrote:I tilt more on the US side like you, but not nearly to your level as I try to get a bit more small/value in international using DLS/DGS.
Yes, over time I may try to even the tilt more between US:Non-US but maintain the overall portfolio factor load targets. There should be no difference in long-term expected return but some in diversification.
RaleighStClaire wrote:Question: How has Schwab been performing on their RAFI funds in comparison to Powershares?
Don’t know for sure, will check, but last time I looked there was a fairly big difference in the numbers of stocks held in each fund and the relative performance. Not sure on the cause of the difference.
bhzmark wrote:But wouldn't it be more useful to just compare the equity allocations as 100% equity allocations?
And fixed income assets are a different comparison and analysis?
Like Rodc, I prefer to look at a portfolio as a whole, as that is how I invest 75:25 stock:bond allocation. However, the analysis presented above reflects the size, and value loads from the FF three factor equity model, and the term loads on the FF two factor fixed income model.
Lucio wrote:How do you estimate the factor loads for the international equity portfolios? Isn't it problematic to use a benchmark series that incorporates multiple markets, currencies and tax regimes?
Here is how I derived the estimates: http://www.bogleheads.org/forum/viewtop ... ght=#96280
And an example of the results for EAFE Value:

Code: Select all

                  Alpha     Beta     Size     Value        R^2 
EAFE Value        
Coefficient       -0.08     0.97    -0.07      0.32       0.97 
t-statistic       -1.20    57.60    -2.53     12.45
In an earlier analysis (for a slightly different time period) DFA estimated the size and value load on EAFE Value to be -0.05 and 0.38 respectively. Not too far off my estimate in the table above. See pg. 10: http://www.ifa.com/Media/Images/PDF%20f ... alysis.pdf The value load I used in the earlier calculation was 0.35.

Robert
.
User avatar
ddb
Posts: 5511
Joined: Mon Feb 26, 2007 11:37 am
Location: American Gardens Building, West 81st St.

Post by ddb »

Robert, thanks for the analysis.

Have you considered reducing overall equity exposure while bumping up HmL and SmB exposure to maintain the same expected return with greater fixed income holdings? In other words, what made you initially commit to the arbitrary 0.2/0.4 factor loadings?

- DDB
"We have to encourage a return to traditional moral values. Most importantly, we have to promote general social concern, and less materialism in young people." - PB
caklim00
Posts: 2420
Joined: Mon May 26, 2008 10:09 am

Post by caklim00 »

ddb wrote:Robert, thanks for the analysis.

Have you considered reducing overall equity exposure while bumping up HmL and SmB exposure to maintain the same expected return with greater fixed income holdings? In other words, what made you initially commit to the arbitrary 0.2/0.4 factor loadings?

- DDB
I had a thread devoted to this very topic: http://www.bogleheads.org/forum/viewtop ... sc&start=0

And, I think Robert gave some insight into the strategy he used when devising the 1,.2,.4 target for equity along with 25% bonds...
Lucio
Posts: 286
Joined: Wed Apr 08, 2009 10:31 pm

Post by Lucio »

Robert T wrote: And an example of the results for EAFE Value:

Code: Select all

                  Alpha     Beta     Size     Value        R^2 
EAFE Value        
Coefficient       -0.08     0.97    -0.07      0.32       0.97 
t-statistic       -1.20    57.60    -2.53     12.45
In an earlier analysis (for a slightly different time period) DFA estimated the size and value load on EAFE Value to be -0.05 and 0.38 respectively. Not too far off my estimate in the table above.
Robert
.
Wow. I am really impressed that the model works that well for such a disparate group of markets. Thanks for that, Robert.

Lucio
User avatar
Random Musings
Posts: 6756
Joined: Thu Feb 22, 2007 3:24 pm
Location: Pennsylvania

Post by Random Musings »

bhzmark wrote:
Thanks for sharing the info. But wouldn't it be more useful to just compare the equity allocations as 100% equity allocations?

And fixed income assets are a different comparison and analysis?

In any case, thanks for doing the work and sharing it.
Rodc wrote:

My preference is to include FI, as that is what I do in real life. There is an interplay between equity and FI and I like it in the evaluation.

There may be value in looking at things such as stocks in isolation from other factors if one is trying to understand how stock markets work. But as an investor, I prefer to think about things at the portfolio level.
I would second the equity only look comparison - to see how those compare. That's 100/0 allocation. Then perhaps a 50/50 and 25/75 to see how the portfolio acts in the whole when at 25, 50, 75 and 100% equities.

Thanks to Robert T for all the work.

RM
User avatar
Topic Author
Robert T
Posts: 2803
Joined: Tue Feb 27, 2007 8:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Post by Robert T »

.
ddb wrote:Have you considered reducing overall equity exposure while bumping up HmL and SmB exposure to maintain the same expected return with greater fixed income holdings? In other words, what made you initially commit to the arbitrary 0.2/0.4 factor loadings?
Random Musings wrote:I would second the equity only look comparison - to see how those compare. That's 100/0 allocation. Then perhaps a 50/50 and 25/75 to see how the portfolio acts in the whole when at 25, 50, 75 and 100% equities.
  • If the equity and fixed income factor loads are exactly the same there is no expected return difference between any combinations. That is what the FF model says. May run some numbers later.
Robert
.
the fixer
Posts: 75
Joined: Sat Jul 25, 2009 6:19 pm

Re: Performance update - Vanguard, iShares, DFA, RAFI

Post by the fixer »

Robert T wrote:.
FWIW I have just updated...
A few things going on here for those interested in another view:

1. In a previous post I showed that when you compare the various DFA and retail indexes side by side over the longest period available, the ETF portfolio actually has a noticeably stronger value tilt (I reran the numbers through year end for highest accuracy and came to a similar conclusion). So a 0.1% advantage for the ETF portfolio (the actual DFA TM portfolio's returns were +11.1% from 03-09) advantage shouldn't be a huge surprise given the noticeably stronger value tilts.

For example, simply using US Marketwide Value, Targeted Value (60/40), and Int'l Value, and Int'l Small (60/40) actually gets you closer tilts to the ETF allocation than the version shown above that includes US Small and Int'l Large Cap. This version outpaced the ETF allocation by about 0.5% per year, FYI.

2. The ETF portfolio benefited from concentrated exposure to mid cap stocks (and the highest performing midcap index of the many now available), which strongly outperformed both large and small value over this period. But before you mistakenly assume that midcap value stocks have some long term expected superiority over large or small, do realize this period was simply reversion to the mean after a lousy 86-99 (incept. for Russell Mid Value--the oldest midvalue index):

86-99

DFA Large Value Index = +17.5%
DFA Marketwide Value Index = +17.3%
Russell Mid Value Index = +14.3%
DFA Small Value Index = +14.6%

Clearly, after this dreadful 15 year stretch, midcap had some catching up to do--and we saw it during this period. Actually, the huge midcap value premium (relative to Marketwide Value) is the entire reason the ETF portfolio beat the less tilted DFA allocation. Every other ETF underperformed or matched its DFA fund counterpart.

Will this run for mid cap continue?

3. Maybe a case of survivorship bias going on here? I seem to remember the original two portfolios (unlike, in my view, the mismatched DFA TM and ETF allocations) designed with similar expected returns were the ETF allocation and the DFA Vector Portfolio. IIRC, the TM allocation came around after the Vector version started seriously pulling away in live results. After an almost 5% advantage in 2009, the Vector allocation's returns since '03 (the actual was 12.2%) comes in at 1% north of the ETF strategy. Not 2% or 3%, but not 0% either. And this comes in before rebalancing costs, which aren't likely to be equal for a ETF allocation with double the funds.

All in all, each allocation looks reasonable, if not a bit clunky (esp. the DFA TM mix--which is just a "tracking portfolio", no one with DFA access would actually choose to invest that way). Too many question marks surrounding the RFAI strategies at this time, IMO. And I'm not sold on using factor analysis as your only allocation tool (if for no other reason that the lack of precision or potential input errors), esp. given the lack of longer term data on most retail indexes.
User avatar
Topic Author
Robert T
Posts: 2803
Joined: Tue Feb 27, 2007 8:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Post by Robert T »

.
the fixer,

I’m not going to go around again on debating differences in the factor load estimates and portfolio comparisons on each of your points. You and I just get different results – that's the difference, simple as that. I believe mine more, you believe yours more. No problem. My suggestion to others is do you own independent analysis, come to your own conclusions.

Robert

PS: If you redo the DFA returns based on the decimaled allocations used in the factor load analysis (which were rounded up in the table, as with the ETF portfolio), I am sure you will get the 11.0% return, as presented in my earlier table (DFA US TM SC = 13.5%, DFA EM = 9.5).
the fixer
Posts: 75
Joined: Sat Jul 25, 2009 6:19 pm

Post by the fixer »

Robert T wrote:.
the fixer,

I’m not going to go around again on debating differences in the factor load estimates and portfolio comparisons on each of your points. You and I just get different results – that's the difference, simple as that. I believe mine more, you believe yours more. No problem. My suggestion to others is do you own independent analysis, come to your own conclusions.

Robert

PS: If you redo the DFA returns based on the decimaled allocations used in the factor load analysis (which were rounded up in the table, as with the ETF portfolio), I am sure you will get the 11.0% return, as presented in my earlier table (DFA US TM SC = 13.5%, DFA EM = 9.5).
I totally agree, and think everyone benefits from hearing different views, right or wrong. That is why I said I was simply offering another view...not necessarily the right one or wrong one.
Post Reply