MidCap Value Tax Efficiency

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
Topic Author
Robert T
Posts: 2574
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

MidCap Value Tax Efficiency

Post by Robert T » Wed Dec 05, 2007 8:54 am

.
Have midcap value ETFs been less tax efficient than their large and small cap value counterparts?

Its often argued the mid-caps are less tax efficient as they have turnover and both end – ie. from mid to large and from small to mid. To try to answer this question I looked at two sets of performance numbers for the iShares Russell and S&P series respectively as they both have performance numbers for the past five years.

For the Russell series
  • - Turnover of the midcap value ETF was much lower than for small value.
    - The tax cost ratio for the midcap value ETF was lower than both large value and small value (although for the latter only marginally so) – perhaps due to its relatively lower dividend yield and relatively high QDI.
    - The % underperformance relative to the corresponding index was lowest for midcap value – relative to large and small cap value and similar to large cap market (ie. Russell 1000).
    - So it seems the iShares Russell Midcap value has been relatively tax efficient.
For the S&P series
  • - Turnover of the midcap value ETF was almost directly in the middle of the turnover for large and small cap value.
    - The tax cost ratio for the midcap value ETF was higher than the small value ETF, but lower than large value).
    - It is quite striking that the iShares S&P600 has been the most tax efficient of the 4 S&P series included in the table below (more so than the iShares 500, 500 value and 400 value ETFs). It has low dividend yield and high QDI.
    - The % underperformance of the corresponding index was lowest for small cap value – relative to large and mid cap value – although only marginally so for the latter.
    - So it seems the iShares S&P 400 value has been relatively tax efficient (relative to the iShares S&P500 and S&P500 value), but not more so than the iShares S&P600 value.

Code: Select all

                            iShares Russell	                iShares S&P
                        1000  1000v MidCapV 2000v      500   500v   400v   600v
Tikker                   IWB   IWD    IWS    IWN       IVV   IVW    IJJ    IJS
Turnover	 	 	 	  	 	 	 	 	 
2003                      5     20     24     45        5     22     11     14
2004                      5     12     11     16        2      5     11     12
2005                      5     15     20     23        6      5     10     13
2006                      7      7     10     14        7      7     21     16
2007                      7     14     24     36        5     20     21     28
Average 2003-07           6     14     18     27        5     12     15     17
 	 	 	 	 	 	 	 	 	 
Tax-cost ratio         0.35   0.47   0.42   0.43     0.36   0.40   0.34   0.27
 	 	 	 	 	 	 	 	 	 
Dividend Yield          1.7    2.5    2.2    2.3      1.8    1.4    2.0    1.5	 	  	 	 	 	 	 	 	 
QDI 2006                100    100     89     69      100    100     88     94
 	 	  	 	 	 	 	 	 	 
5yrs to end Sept 2005	 	 	 	
 	 	 	 	 	 
Index Return           16.0   18.1   21.0   18.7     15.5   18.2   19.6   18.4
After-tax Return       15.4   17.3   20.3   17.9     14.9   17.5   18.9   17.8
Difference             -0.6   -0.8   -0.8   -0.8     -0.5   -0.7   -0.7   -0.6  
% index return lost    -3.5   -4.3   -3.6   -4.2     -3.6   -3.8   -3.4   -3.3

Source: Derived from information on the iShares website.
Portfolios with midcaps

Assuming the above relative tax efficiency continues, IMO a midcap value:microcap combination does a reasonable job at targeting value and size exposure. Consider the two portfolios below, by my best estimate, both have similar size and value loads – around 0.3 and 0.4 respectively (the size load is slightly higher on P2 than P1 – closer to 0.4).

P1
25% Vanguard TSM
50% iShares Russell MidCap Value
25% Bridgeway Ultra Small Company Market

P2
50% Vanguard TSM
50% DFA Small Value

Code: Select all

2002-2006

       Annualized    Standard
         Return     Deviation    Sharpe
P1       15.24        18.53       0.688
P2       13.21        19.97       0.547
The reason for the differing returns of portfolios with similar factor loadings, IMO is due mainly to the non-linearity in the size premium over this period - CRSP10 outperformed other deciles by a wide margin. Its likely that P1 has a higher exposure to microcaps than P2 which is not picked up in the factor load estimates. So P1 may have slightly higher risk than P2 despite having similar factor loads.

Using data from 1996-2006 for some of the indexes

P1
25% TSM
50% Russell MidCap Value
25% CRSP10

P2
50% TSM
50% DFA Small Value

Code: Select all

1996-2006

       Annualized    Standard
         Return     Deviation    Sharpe
P1       14.14        15.90      0.717      
P2       13.42        16.20      0.665  
For the 1996-2006 period, P1 doesn’t include any costs, while P2 includes them for the DFA small value fund.

IMO the main reason for the historical outperformance of the DFA small value fund is due to its higher size loading relative to comparators (i.e. it includes microcaps which have performed well relative to other size deciles over recent years). If a portfolio is constructed to get similar microcap exposure then IMO returns will be similar. This can be done, arguable with more control, with a mid-cap value and micro-cap combination (as many small value funds don't have exposure to micro-caps). For example the median market cap of the DFA small value fund is $171m, while for the Russell 2000 value it is $615m.

So from the above, IMO a Russell midcap value:Bridgeway ultra-small company market combination is a reasonable way to target value and size exposure. The resulting performance will likely get very close to the returns of some of the DFA fund combinations (for those targeting US value tilts of about 0.5-0.6 and under). It would have been good to assess the tax efficiency of the MSCI midcap value but we don’t yet have much data.

Robert
.

SmallHi
Posts: 1718
Joined: Wed Feb 21, 2007 6:11 pm

Post by SmallHi » Wed Dec 05, 2007 12:05 pm

Robert,

Thanks for the analysis. This has indeed been an interesting 5-10 years. I agree with just about everything you say...

Thoughts

Mid Cap Value ETFs should continue to be very tax efficient and suitable for taxable accounts. I have a marginal preference for MSCI 450 Value or Russell Mid Value over S&P 400 Value due to higher value loading and (?) slightly higher security count, but it doesn't matter much.

And, ignoring a bit of negative tracking error recently, BRSIX has done a fine job of capturing the micro cap market returns tax efficiently.

As we have talked about before, I would expect a more pronounced size effect in CRSP 10 than CRSP 1-9 going forward due to the liquidity risks, although they ain't what they used to be. (Dimensional estimated that bid/asked spreads in the microcap market have dropped from about 18% in the early 90s to under 2% today).

Concerns

I guess my biggest point of "skepticism" is whether or not Mid Cap Value will continue to be the sweet spot on the value spectrum going forward?

Part of why the MCV/Micro mixes work so well relative to TSM/SV is because MCV has outpaced LV and SV by a wide marging over the last 10, 15, and 20 years!

Code: Select all

1986-2007

1000 Value = +12.9%
Mid Value =  +14.0%
FF SV xU =   +13.9%
And, as much as you want to say "hey, maybe this is a Russell phenominon", that isn't the case. Dimensionals US Vector portfolio would have actually exceeded the Russell Mid Value's returns since 1986 by about 0.1% or so (despite a higher size loading and lower value loading -- but a hefty overall mid cap committment of around 40% or so). So, anything mid cap oriented in nature has had the upper hand since at least the mid 1980s.

Alternative time frames

But if we look back to the 20 years prior to 1986, we find a very linear relationship between value co-horts:

Code: Select all

1965-1985

FF LV =    +13.9%
Vector** = +14.3%
FF SV =    +18.1%

CRSP 1-2 =   +8.2%
CRSP 3-5 =  +12.4%
CRSP 6-8 =  +14.4%
CRSP 9-10 = +14.4%

** admitedly, Vector is a poor proxy for Mid Value, as its not as tilted to HmL as LV or SV is (although using DFA Marketwide Value in place of US Vector would have produced a similar return deficit relative to USSV -- and Mktwide Value has had a similar size/value tilt to MCV since 1986)
So, it may very well be that the last 20 years for Mid Value and CRSP 10 were a "reversion to the mean" if you will to correct for a difficult prior 2 decades.

My beliefs

I am still more of a believer of "going light" on CRSP 1 and mega growth than I am "overweigting" any particular size or value decile. Simply put, I think more returns are to be had relative to the market by skipping the mega cap, high P/E stocks than extreme tilts to a particular dimension (say CRSP 10 or MCV). Both would be the optimal solution, but its tough to do so without extreme factor loadings.

I would not be uncomfortable with a Mid Value/CRSP 10 approach to size and value tilting, I guess I would just feel more comfortable adding Small Value to the fray -- a modified Mid Value, Small Value/CRSP 10 allocation.

Conceptually, I prefer the idea of obtaining risk factor tilts in as broad a fashion as possible. For example, for those with DFA access, I would prefer to gain value exposure with a combo of US LV and SV or Marketwide Value and Targeted Value (each set owns every stock in the value universe). If you plan calls for something in the 0.4/0.4 range, my preference would be for US Vector.

For those without DFA, I would stick with Mid/Small Value and a Small or Micro Cap market fund...although I may just decide to increase my value tilt relative to my size tilt and exclude Micro Caps altogether -- using just TSM, Mid Value, and Small Value.

Thanks again Robert!

SH (not medium high :lol: )

User avatar
Topic Author
Robert T
Posts: 2574
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Post by Robert T » Wed Dec 05, 2007 1:53 pm

.
SH,

1. I also slightly prefer the MSCI and Russell mid cap value over the S&P 400v due to their higher R^2, lower negative alpha, no dramatic difference in tax efficiency and to some extent lower size loading (in the FF3F regressions). The reverse is true for small value. (FWIW I still use S&P400v).

2. Yes I agree, the higher recent MCV and especially micro may just be revision to the mean – but I think it is revision to the mean of the size dimension not value (size premium has been more volatile than the value premium).

3. If we look at the size decile returns over the 1986-2007 period – they are fairly constant, bar the first decline and with similar value premiums across the size spectrum, the returns were similar – as in your table.

4. On the 1965-85 period, I think the size effect is again influencing return differences rather then value premium (which explains the linear relationship between the value cohorts).

Code: Select all

From your table                 
                     
LV                         13.9  
SV                         18.1  
Difference                 +4.1

Assume each of the above get their size exposure from:

(CRSP1-2 + CRSP3-5)/2      10.3   
CRSP9-10                   14.4
Difference                 +4.2
  • So one could argue that differences in size exposure drove the return differences between LV and SV (as the value premium is fairly constant across LV and SV).

    For example – using the above (a rough estimate)

    Large ‘value premium’ = 13.0 – 10.3 = 3.6
    Small ‘value premium’ = 18.1 - 14.4 = 3.7
5. I agree, at the end of the day comfort level matters. So far, I have been more comfortable spreading my value and size exposure across the size and value spectrum – however this may come at the expense of lagging a corresponding FF factor benchmark (which seems more likely to be matched with the MCV/CRSP10 combination – due, as far as I can tell, to the non-linearity in the size premium - disproportionately higher in CRSP10 due to disproportionately higher risk than the other deciles – and that the factor benchmarks capture some of this non-linearity - I could be wrong).

Robert
.

vicrana
Posts: 59
Joined: Fri Jul 27, 2007 9:58 am

Post by vicrana » Thu Jan 03, 2008 11:09 am

How does Vanguard Mid cap Index or Value sound for Mid cap exposure?

User avatar
Topic Author
Robert T
Posts: 2574
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Post by Robert T » Thu Jan 03, 2008 4:02 pm

.
Vicrana,
How does Vanguard Mid Cap Index or Value sound for Mid cap exposure?
IMO there are three main criteria for security selection:
  • (i) Consistent risk exposure to minimize tracking error relative to risk allocation targets
    (ii) Tax efficient and low cost to minimize the difference between actual and benchmark returns. As tax rates are likely to increase faster than expense ratios for taxable accounts taxes should be given priority.
    (iii) Reliable fund stewardship to ensure consistent exposure to compensated risk and continuous compounding.
Following these three criteria:

Consistent risk exposure:
  • Factor exposure: The back test data for the midcap value index that the Vanguard fund tracks looks good (i.e. captured about three quarters of the value premium and a small share of the size premium, and is explained fairly well by the three factor model with low and insignificant negative alpha).

    Tracking error: Vanguard index funds (pre-tax) seem to track their indexes more closely than iShares index funds, at least for the funds I have looked at [more closely than the expense ratio difference suggests]. I would expect the same would be true for the midcap series.
Tax-efficient and low cost:
  • Expense ratio: They have the lowest expense ratios of all midcap index funds (if I am not mistaken).

    Tax-efficiency: The greatest unknown going forward IMO is the level of tax-efficiency. It will likely be less tax-efficient than the iShares Midcap value ETFs (because of structure differences and due to its slightly higher REIT allocation (about 12%) relative to the iShares Russell Midcap value and iShares S&P400 value (about 8%)). Whether its closer pre-tax tracking of the underlying index will offset its lower tax-efficiency to give smaller after-tax return differences with the underlying indexes than the iShares funds is unclear, but think it will be fairly close.
Reliable fund stewardship:
  • Vanguard is one of the best, and IMO currently slightly more reliable than BGI.
Based on the above - I currently think the Vanguard MidCap Value fund is a good/reasonable choice for midcap value exposure (we will need more years of data to determine its tax-efficiency).

Robert

Edited to hopefully provide a clearer answer.
.

User avatar
bzboy
Posts: 252
Joined: Fri Mar 02, 2007 5:46 am

Post by bzboy » Thu Jan 03, 2008 4:33 pm

Robert T wrote:.
1. I also slightly prefer the MSCI and Russell mid cap value over the S&P 400v due to their higher R^2, lower negative alpha, no dramatic difference in tax efficiency and to some extent lower size loading (in the FF3F regressions). The reverse is true for small value. (FWIW I still use S&P400v).
.
I have been a fan of MV and have been an investor in IWS for several years now. For my new money, I am inclined to go to Morningstar Mid Value (JKI) as I think that's more valuey. I would appreciate if you throw that into the mix.

bz

User avatar
Topic Author
Robert T
Posts: 2574
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Post by Robert T » Thu Jan 03, 2008 10:36 pm

I am inclined to go to Morningstar Mid Value (JKI) as I think that's more valuey. I would appreciate if you throw that into the mix.

Code: Select all

Annualized Returns as at 12/31/2007

Index                                1 yr   3 yrs   5 yrs   10 yrs

S&P MidCap 400/Citigroup Value       2.65    9.23   15.36     8.52
Russell MidCap Value Index          -1.42   10.11   17.92    10.18
MSCI MidCap Value                   -4.41    8.32   17.39    10.77
Morningstar MidCap Value            -5.52    7.78   16.17     9.41

Source: iShares and MSCI websites.
There was quite are large return spread in 2007 (about 8%) between the best performer and worst performer of the above midcap value indexes. My hypothesis is that both the MSCI and Morningstar index had a higher allocation to REITs and hence did relatively poorly. The S&P and Russell index have a similar allocation to REITs (about 8%) so I think the difference in performance between these two indexes is the difference in value loadings (smaller for the S&P index and as the value ‘premium’ was negative last year it had slightly higher return). The ishares fund tracking the Morningstar index had a higher tax-cost ratio than the ETFs tracking the S&P and Russell indexes (i.e. was least tax efficient) over the last three years, perhaps a reflection of a higher REIT allocation.

My sense from the above returns is that the Morningstar series is not more value oriented than the others (I could be wrong and will test with the 3 factor model when I have more time). The P/B comparisons from M* and the iShares websites seem to give a contradictory story.

Robert
.
Last edited by Robert T on Sun Mar 09, 2008 2:19 am, edited 1 time in total.

User avatar
bzboy
Posts: 252
Joined: Fri Mar 02, 2007 5:46 am

Post by bzboy » Fri Jan 04, 2008 2:09 pm

Robert T wrote: My sense from the above returns is that the Morningstar series is not more value oriented than the others (I could be wrong and will test with the 3 factor model when I have more time).
.
Thanks, Robert. M* value is actually the bottom third of the market (more like DFA and less like S&P, Russell and MSCI) though they differ on how they determine value. It's possible that M* value includes a bit more of Homebuilders (they are deep value now) and hence the wide spread.

I will stay tuned to your 3 factor analysis, whenever you get to it.

Thanks,
bz

stan1
Posts: 7114
Joined: Mon Oct 08, 2007 4:35 pm

Post by stan1 » Fri Jan 04, 2008 2:39 pm

Just for completeness, what about RFV (Rydex S&P400 Pure Value)?

I know there are the usual Rydex concerns: thinly traded, liquidity, bid/ask spread, small number of holdings, uncertain corporate commitment to maintaining low ERs, potential for liquidation, etc.

Treating those as known risks separate from the performance of the index, how does RFV compare to the others?

vicrana
Posts: 59
Joined: Fri Jul 27, 2007 9:58 am

Post by vicrana » Fri Jan 04, 2008 2:53 pm

Robert T wrote:.
Vicrana,
IMO there are three main criteria for security selection:
  • (i) Consistent risk exposure to minimize tracking error relative to risk allocation targets
    (ii) Tax efficient and low cost to minimize the difference between actual and benchmark returns. As tax rates are likely to increase faster than expense ratios for taxable accounts taxes should be given priority.
    (iii) Reliable fund stewardship to ensure consistent exposure to compensated risk and continuous compounding.
Following these three criteria:

Consistent risk exposure:
  • Factor exposure: The back test data for the midcap value index that the Vanguard fund tracks looks good (i.e. captured about three quarters of the value premium and a small share of the size premium, and is explained fairly well by the three factor model with low and insignificant negative alpha).

    Tracking error: Vanguard index funds (pre-tax) seem to track their indexes more closely than iShares index funds, at least for the funds I have looked at [more closely than the expense ratio difference suggests]. I would expect the same would be true for the midcap series.
Tax-efficient and low cost:
  • Expense ratio: They have the lowest expense ratios of all midcap index funds (if I am not mistaken).

    Tax-efficiency: The greatest unknown going forward IMO is the level of tax-efficiency. It will likely be less tax-efficient than the iShares Midcap value ETFs (because of structure differences and due to its slightly higher REIT allocation (about 12%) relative to the iShares Russell Midcap value and iShares S&P400 value (about 8%)). Whether its closer pre-tax tracking of the underlying index will offset its lower tax-efficiency to give smaller after-tax return differences with the underlying indexes than the iShares funds is unclear, but think it will be fairly close.
Reliable fund stewardship:
  • Vanguard is one of the best, and IMO currently slightly more reliable than BGI.
Based on the above - I currently think the Vanguard MidCap Value fund is a good/reasonable choice for midcap value exposure (we will need more years of data to determine its tax-efficiency).

Robert

Edited to hopefully provide a clearer answer.
.
Thanks Robert for your detailed analysis.

User avatar
Topic Author
Robert T
Posts: 2574
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Post by Robert T » Fri Jan 04, 2008 2:53 pm

.
Bz,

You are right.

Here are the estimated factor loads using the available data I have. The Morningstar MidCap value series has the highest value loading (together with the S&P pure value series) of the five indexes below. For some reason it also has the lowest beta so its expected returns may not be much higher than the MSCI and Russell series. An interestingly result nevertheless. The S&P pure value series has the same value load as the Monringstar series, a similar low beta, but it has a larger small cap tilt. The 3F model does not explain the S&P pure value series as well as the others (R^2 of 0.8.).

Code: Select all

Three factor regression results using data from July 1997-Sept 2006.
                                                                          
                                   Alpha        Beta    Size   Value     R^2                             
Index                          Coeff.(t-stat)                                

MSCI US Mid Cap Value Index    -0.01   -0.08    1.04    0.11    0.79    0.89  
Morningstar MidCap Value       -0.13   -0.84    0.97    0.04    0.89    0.87  
Russell Mid Cap Value          -0.06   -0.51    1.02    0.09    0.74    0.91  
S&P 400/Citigroup Value        -0.19   -1.15    1.04    0.21    0.63    0.87
S&P 400/Citigroup Pure Value   -0.18   -1.11    0.97    0.33    0.89    0.80 

Robert
.

User avatar
bzboy
Posts: 252
Joined: Fri Mar 02, 2007 5:46 am

Post by bzboy » Fri Jan 04, 2008 11:31 pm

Robert T wrote:.
For some reason it also has the lowest beta so its expected returns may not be much higher than the MSCI and Russell series.
Thank you, Robert. Appreciate the info.

bz

User avatar
tc101
Posts: 3108
Joined: Tue Feb 20, 2007 3:18 pm
Location: Atlanta - Retired in 2004 at age 54

Post by tc101 » Fri Mar 07, 2008 2:58 pm

This conversation has been about mid cap value ETFs. I have a related question that is probably better to put here than in a whole new topic.

What about the mid cap index fund? Not value but the total mid cap index mutual fund? I know it won't be as tax efficient as the total stock market index fund or the tax managed small cap fund which are what I currently have in my taxable account, but will it be fairly tax efficient?

I have read other discussions here that have convinced me that having a portion of my USA allocation in mid cap is a good idea, but the thing I am still trying to figure out is if it is a good idea in taxable, or would I be better off just splitting the money between total stock market index fund and tax managed small cap fund and forgetting about mid cap because of the tax situation?

I am retired and in a pretty low tax bracket, but I still want to do things as efficiently as possible.

Post Reply