An index isn't an index isn't an index...

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
siamond
Posts: 4329
Joined: Mon May 28, 2012 5:50 am

An index isn't an index isn't an index...

Post by siamond » Tue Jan 17, 2017 6:22 pm

Let me advertise another blog post from Barry, a short & sweet and especially striking post, actually.

It provides a table comparing the 2016 returns of various types of indices (the usual 3x3 categorization) per index provider (CRSP, Russell, etc). For sure, there are differences of methodology, but some of the differences are a little bit eye-popping. If somebody has a good explanation for the fact that CRSP seems pretty poor in comparison for *every* category, I'd love to hear it. :shock:

https://blbarnitz4.wordpress.com/2017/0 ... x-returns/

This shows really well that index investing is great, but it's even better with solid indices...

Theoretical
Posts: 1440
Joined: Tue Aug 19, 2014 10:09 pm

Re: An index isn't an index isn't an index...

Post by Theoretical » Tue Jan 17, 2017 6:32 pm

Ouch. That's a terrible showing all around.

The really embarrassing one to me is that S&P small growth was only 3 percentage points off CSRP small value in a year when small value beat small growth by almost 16 percentage points.

User avatar
grabiner
Advisory Board
Posts: 23149
Joined: Tue Feb 20, 2007 11:58 pm
Location: Columbia, MD

Re: An index isn't an index isn't an index...

Post by grabiner » Tue Jan 17, 2017 9:25 pm

The S&P indexes have a fundamental difference from the others; they are not intended to reflect the whole market, but a subset of companies which meet S&P's criteria. In large-cap, this makes little difference, as almost all large companies are in the S&P 500. But in mid-cap and small-cap, the S&P may not be a representative sample; the S&P 600 and Russell 2000 both cover about the same cap range, but the Russell has many more stocks. This was most noticeable during the Internet bubble; the S&P 600 avoided the bubble, and thus underperformed most small-cap funds in 1997-1999 but outperformed in 2000-2002 when the bubble burst.

Is the Russell 2000 still being front-run regularly? Vanguard's Small-Cap Index, when it tracked the Russell 2000, outperformed the index regularly. The ETF VTWO has only matched the index, although that is still a slight outperformance since it has 0.15% expenses.
Wiki David Grabiner

Dead Man Walking
Posts: 731
Joined: Wed Nov 07, 2007 6:51 pm

Re: An index isn't an index isn't an index...

Post by Dead Man Walking » Tue Jan 17, 2017 11:32 pm

grabiner wrote:The S&P indexes have a fundamental difference from the others; they are not intended to reflect the whole market, but a subset of companies which meet S&P's criteria. In large-cap, this makes little difference, as almost all large companies are in the S&P 500. But in mid-cap and small-cap, the S&P may not be a representative sample; the S&P 600 and Russell 2000 both cover about the same cap range, but the Russell has many more stocks. This was most noticeable during the Internet bubble; the S&P 600 avoided the bubble, and thus underperformed most small-cap funds in 1997-1999 but outperformed in 2000-2002 when the bubble burst.

Is the Russell 2000 still being front-run regularly? Vanguard's Small-Cap Index, when it tracked the Russell 2000, outperformed the index regularly. The ETF VTWO has only matched the index, although that is still a slight outperformance since it has 0.15% expenses.
I think the S&P 600 stock selection requires that the companies have a history of profitability.

DMW

Morik
Posts: 709
Joined: Tue Nov 25, 2014 12:26 pm

Re: An index isn't an index isn't an index...

Post by Morik » Tue Jan 17, 2017 11:56 pm

grabiner wrote:The S&P indexes have a fundamental difference from the others; they are not intended to reflect the whole market, but a subset of companies which meet S&P's criteria. In large-cap, this makes little difference, as almost all large companies are in the S&P 500. But in mid-cap and small-cap, the S&P may not be a representative sample; the S&P 600 and Russell 2000 both cover about the same cap range, but the Russell has many more stocks. This was most noticeable during the Internet bubble; the S&P 600 avoided the bubble, and thus underperformed most small-cap funds in 1997-1999 but outperformed in 2000-2002 when the bubble burst.

Is the Russell 2000 still being front-run regularly? Vanguard's Small-Cap Index, when it tracked the Russell 2000, outperformed the index regularly. The ETF VTWO has only matched the index, although that is still a slight outperformance since it has 0.15% expenses.
Can some of that out-performance be from securities lending that Vanguard does? I know it contributes enough to offset or nearly offset the fees in some of their (very low cost) funds.

*3!4!/5!
Posts: 1256
Joined: Fri Sep 02, 2016 1:47 pm

Re: An index isn't an index isn't an index...

Post by *3!4!/5! » Wed Jan 18, 2017 12:33 am

I'm really only interested in the "Total Market" indices. What explain the differences there?

User avatar
grabiner
Advisory Board
Posts: 23149
Joined: Tue Feb 20, 2007 11:58 pm
Location: Columbia, MD

Re: An index isn't an index isn't an index...

Post by grabiner » Wed Jan 18, 2017 9:53 am

*3!4!/5! wrote:I'm really only interested in the "Total Market" indices. What explain the differences there?
Even there, the construction rules are different; for example, when are new issues added? If a large new stock rises or falls between IPO and index addition, it matters which indexes contained it.
Wiki David Grabiner

User avatar
House Blend
Posts: 4517
Joined: Fri May 04, 2007 1:02 pm

Re: An index isn't an index isn't an index...

Post by House Blend » Wed Jan 18, 2017 10:05 am

FWIW, there are some good pages in the wiki that contain far more historical and current comparative information about indexes. Unfortunately they seem well-hidden. Here's the one with history for total US market indexes:
https://www.bogleheads.org/wiki/US_tota ... ex_returns

I'm too lazy to look under the hood, but not too lazy to speculate that Barry is the one who created those pages.
grabiner wrote:Is the Russell 2000 still being front-run regularly? Vanguard's Small-Cap Index, when it tracked the Russell 2000, outperformed the index regularly. The ETF VTWO has only matched the index, although that is still a slight outperformance since it has 0.15% expenses.
FWIW #2: I have an R2000 index fund (TISBX) in my employer's plan that has a very low ER--currently 0.06%. It has been able to outperform the index by ~(1/4)%/yr over the past 5 years. The fund does do some securities lending, so that may explain some of the outperformance. AFAIR, back in the day when the VG small cap index was benchmarked against the R2000, it was routinely beating it by much larger amounts.

Code: Select all

      TISBX   R2000   delta
2016 21.58%  21.32%  +0.26%
2015 -4.14%  -4.41%  +0.27%
2014  5.15%   4.89%  +0.26%
2013 38.98%  38.82%  +0.17%
2012 16.61%  16.33%  +0.28%

Angst
Posts: 1874
Joined: Sat Jun 09, 2007 11:31 am

Re: An index isn't an index isn't an index...

Post by Angst » Wed Jan 18, 2017 10:08 am

Bump...
I can accept, somewhat blindly albeit, that the S&P selection process makes comparison with its indexes a little problematic, but the others? I'm curious what's going on. My emphasis below.
siamond wrote:Let me advertise another blog post from Barry, a short & sweet and especially striking post, actually.

It provides a table comparing the 2016 returns of various types of indices (the usual 3x3 categorization) per index provider (CRSP, Russell, etc). For sure, there are differences of methodology, but some of the differences are a little bit eye-popping. If somebody has a good explanation for the fact that CRSP seems pretty poor in comparison for *every* category, I'd love to hear it. :shock:

https://blbarnitz4.wordpress.com/2017/0 ... x-returns/

This shows really well that index investing is great, but it's even better with solid indices...

User avatar
JamesSFO
Posts: 3121
Joined: Thu Apr 26, 2012 10:16 pm

Re: An index isn't an index isn't an index...

Post by JamesSFO » Wed Jan 18, 2017 10:50 am

I think this just shows different methodologies. Not sure that there is more to be said than that. Without studying the index composition methodology it's hard to make other conclusions. For example, S&P's methodology is posted https://us.spindices.com/documents/meth ... ndices.pdf so then you have to read the full methodology and compare it to CRSP's approach http://www.crsp.com/files/Equity-Indexe ... uide_0.pdf to understand the difference.

The S&P SmallCap 600 says " US$ 400 million to US$ 1.8 billion for the S&P SmallCap 600" while CRSP says "Index range: 85% < Xi,t ≤ 98%". Harder to directly translate but my guess is that CRSP is tilting both larger and smaller and quick checks of the constituent list finding larger market cap companies than the S&P range (I found 3.53B market cap companies and $182M market cap companies).

So the two indices are not holding comparable things and thus aren't really comparable despite the naming.

Theoretical
Posts: 1440
Joined: Tue Aug 19, 2014 10:09 pm

Re: An index isn't an index isn't an index...

Post by Theoretical » Wed Jan 18, 2017 11:06 am

I did some factor regressions on Russell 2K and 2K value funds with the time barriers being origins of the fund to the end of year 2006. 2007 was the first time Russell implemented the sizing buffer zones. The difference is drastic, though the Russell 2000 is now only moderately desirable instead of abysmal.

Funds compared: Blackrock Small Cap Index Institutional, iShares Russell 2000 and iShares Russell 2000 Value. The Blackrock fund is the only one going back to 1997 and I couldn't find an R2K value index fund before 2000. Vanguard's followed it, but switched indexes enough it wouldn't be a fair comparison.

Pre-2007: http://bit.ly/2k06jC0

Post-2007: http://bit.ly/2j971Nm

It's a drastic change, where R2K funds still lose 80-100 bp per year, but it's better than the more than 400 bp annual losses from before the

Angst
Posts: 1874
Joined: Sat Jun 09, 2007 11:31 am

Re: An index isn't an index isn't an index...

Post by Angst » Wed Jan 18, 2017 11:31 am

- I don't think it's useful to focus on comparing the CRSP to the S&P indexes - it's far less meaningful as the S&P is constituted by committee.
- But the Russell, MSCI and Dow indexes are each, in aggregate purportedly total market cap-weighted indexes, like CRSP, and broken out by both capitalization and styles. But if you add all those capitalization and style breakouts back into themselves, you should end up with the total market. So, back to the point: Why does the CRSP end up in last place, essentially across the board? Regardless of total or breakouts. Could their "transitioning bands" (or whatever they're called) explain this in some way? I have no idea. The OP didn't appear to either.

User avatar
Barry Barnitz
Wiki Admin
Posts: 3031
Joined: Mon Feb 19, 2007 10:42 pm
Contact:

Re: An index isn't an index isn't an index...

Post by Barry Barnitz » Wed Jan 18, 2017 1:14 pm

Some distinguishing features of CRSP indexes, according to Vanguard---> A closer look at CRSP [link fixed --admin LadyGeek]
Unlike other domestic providers, CRSP doesn’t use a fixed number of names to construct an index. Instead, it ranks every stock by market
capitalization and then forms its indexes based on specific percentages of the market. For example, under CRSP’s approach, the top 70%
of the market represents the mega-cap universe. And the next 15% of the market-cap spectrum represents CRSP’s mid-cap universe.

The number of names in a benchmark will vary from quarter to quarter. But we consider CRSP’s percentage-based approach, which is more common in stock benchmarks outside the United States, an effective strategy. Because the number of stocks in the entire market can change significantly over time, the characteristics of a benchmark’s market coverage can shift even more dramatically with the traditional approach of a fixed number of names.
A likely candidate for return dispersion vs. other providers:
CRSP also has designed a securities migration approach, called packeting, which lowers turnover and minimizes transaction costs. Packeting allows holdings to be shared between two indexes. A company with a changing market capitalization is moved in separate stages or packets to the next index through a shared market-cap band between them. This helps lower turnover in an index by avoiding the need to move a company all at once when its market cap changes.
For turnover confirmation, see
regards,
Image | blb | December Birthday Celebration: Ludwig van Beethoven

User avatar
siamond
Posts: 4329
Joined: Mon May 28, 2012 5:50 am

Re: An index isn't an index isn't an index...

Post by siamond » Wed Jan 18, 2017 1:32 pm

Angst wrote:So, back to the point: Why does the CRSP end up in last place, essentially across the board? Regardless of total or breakouts. Could their "transitioning bands" (or whatever they're called) explain this in some way? I have no idea. The OP didn't appear to either.
You're right. I don't have a clue. And this bothers me a tad. Notably considering that numerous Vanguard primary funds switched to CRSP as their benchmark starting in 2013.

PS. I think this historical information about indices returns is very informational. I am considering adding those data series to the Simba backtesting spreadsheet (not in the primary list of data series, but in the new Data_Misc tab I recently added).
Last edited by siamond on Wed Jan 18, 2017 1:33 pm, edited 1 time in total.

toto238
Posts: 1880
Joined: Wed Feb 05, 2014 2:39 am

Re: An index isn't an index isn't an index...

Post by toto238 » Wed Jan 18, 2017 1:33 pm

It would appear that the border between small and mid, and mid and large, is largely to blame here.

To oversimplify, last year small caps did better than large caps. The smaller the cap, the better it did.

CRSP indices, due to their construction methods, tend to skew larger-cap. So in their mid-cap index fund they include many stocks that S&P or Russell would consider large-cap. In CRSP's small-cap they include many stocks that S&P or Russell would consider mid-cap. And CRSP almost completely ignores stocks they consider micro-cap that S&P and Russell may consider to still be small-cap.

In a year where large-cap outperforms small-cap, we should expect to see the opposite occurring.

ftobin
Posts: 1023
Joined: Fri Mar 20, 2009 3:28 pm

Re: An index isn't an index isn't an index...

Post by ftobin » Wed Jan 18, 2017 1:35 pm

I get the impression that CRSP indexes are designed to a large amount of funds to be invested in them without hurting the index. The more extreme the index (small caps, value-ness), the narrow the available securities and the impact of trading them is higher. The other indexes were designed before the advent of mega index funds, and likely weren't originally intended for a large amount of funds to track them.

jalbert
Posts: 3969
Joined: Fri Apr 10, 2015 12:29 am

Re: An index isn't an index isn't an index...

Post by jalbert » Wed Jan 18, 2017 2:24 pm

The total market index comparisons are the only ones worth exploring. The rest track different market segments. For instance, with large caps, S&P is about 500 stocks, MSCI is 750, and Russell is 1000. CRSP has unconventional naming: their large cap index is a combination of their megacap and midcap index, and currently tracks 609 stocks.

The variation in total market index returns might be a result of slightly different sector weights. For instance, comparing CRSP to S&P, CRSP has slightly lower weightings for tech and healthcare and a slightly higher weighting for energy.
Risk is not a guarantor of return.

User avatar
in_reality
Posts: 4529
Joined: Fri Jul 12, 2013 6:13 am

Re: An index isn't an index isn't an index...

Post by in_reality » Thu Jan 19, 2017 4:38 am

siamond wrote:Let me advertise another blog post from Barry, a short & sweet and especially striking post, actually.

It provides a table comparing the 2016 returns of various types of indices (the usual 3x3 categorization) per index provider (CRSP, Russell, etc). For sure, there are differences of methodology, but some of the differences are a little bit eye-popping. If somebody has a good explanation for the fact that CRSP seems pretty poor in comparison for *every* category, I'd love to hear it. :shock:

https://blbarnitz4.wordpress.com/2017/0 ... x-returns/

This shows really well that index investing is great, but it's even better with solid indices...

The Boglehead wiki shows the 5 and 10 year CAGR of the CRSP is worse than the DOW JONES' and MSCI's. In fact, the CRSP is nearly the worst for the seven indexes given. It's 0.07% and 0.13% worse than the average index performance.

Some of that is backdated data though....

https://www.bogleheads.org/wiki/US_tota ... ex_returns

dcabler
Posts: 696
Joined: Wed Feb 19, 2014 11:30 am

Re: An index isn't an index isn't an index...

Post by dcabler » Thu Jan 19, 2017 7:27 am

Speaking of multiple indices, what's with Vanguard having two MCV ETFs?
VOE - Currently tracks the CRSP MCV index. This is the one that most commonly shows up on Vanguard's site
IVOV - Tracks the S&P MCV 400 Index. You have to dig to find it - it doesn't even show up on the "complete Vanguard ETF" list on their site.

As noted in this thread, the dispersion of returns is pretty high between the two since 2013 when Vanguard switched to CRSP.
IVOV is quite a bit smaller than the equivalent IJJ ($632M vs. $5.97B) and I've seen some articles that state liquidity concerns.

Seems it's been somewhat difficult to find on Vanguard's website for a while - there's an older thread on this forum where people are questioning whether this fund is on the list for eventual shut down. But it's still there....

User avatar
Doc
Posts: 8752
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: An index isn't an index isn't an index...

Post by Doc » Thu Jan 19, 2017 7:46 am

IVOV average trading volume only 32600 shares. Bye bye?
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

animule
Posts: 130
Joined: Sun Jan 06, 2008 12:29 pm

Re: An index isn't an index isn't an index...

Post by animule » Thu Jan 19, 2017 8:17 am

The S&P 600 is a superior index for small caps, mainly because of the earnings requirement (four consecutive quarters of positive earnings) which screened out a lot of internet startups trading at huge valuations with no profits around 2000, just before they cratered and cost investors tons of money.
An additional potential advantage of the S&P 600 over the Russell 2000 is the profitability screen (requiring four consecutive quarters of positive earnings) that is used by its selection committee. We now know that direct profitability is an important dimension of expected returns, and among small caps, growth companies with low or negative profitability have had a much lower return than higher profitability companies, so excluding those companies would provide a large advantage to the S&P 600. Direct Profitability is now one the most important considerations in the funds that IFA advises for our clients.
https://www.ifa.com/articles/Index_Spot ... vs_SP_600/

The only way I am aware of to get this index at Vanguard is through the Tax Managed Small Cap fund.

wolf359
Posts: 1462
Joined: Sun Mar 15, 2015 8:47 am

Re: An index isn't an index isn't an index...

Post by wolf359 » Thu Jan 19, 2017 9:49 am

That is a chart of the performances of the various indices for 2016. That's only one year of data. Each year, the numbers will be different. Different indices will do better or worse in any given year. One year's results doesn't really predict what you'll get next year or over the long term.

It's also not just a matter of picking an index. It's picking a provider. You need someone who can execute well, and actually get close to the performance of the index being tracked.

The other key is the expense ratios of the specific fund you choose. Some indices are more expensive to execute than others.

For example, I have seen a 401K plan with an S&P 500 index fund that charges 0.5%, and it still doesn't do a very good job of tracking that index.

beammeupscotty
Posts: 181
Joined: Thu Jan 21, 2010 8:13 am
Contact:

Re: An index isn't an index isn't an index...

Post by beammeupscotty » Thu Jan 19, 2017 10:07 am

animule wrote:The S&P 600 is a superior index for small caps, mainly because of the earnings requirement (four consecutive quarters of positive earnings) which screened out a lot of internet startups trading at huge valuations with no profits around 2000, just before they cratered and cost investors tons of money.
An additional potential advantage of the S&P 600 over the Russell 2000 is the profitability screen (requiring four consecutive quarters of positive earnings) that is used by its selection committee. We now know that direct profitability is an important dimension of expected returns, and among small caps, growth companies with low or negative profitability have had a much lower return than higher profitability companies, so excluding those companies would provide a large advantage to the S&P 600. Direct Profitability is now one the most important considerations in the funds that IFA advises for our clients.
https://www.ifa.com/articles/Index_Spot ... vs_SP_600/

The only way I am aware of to get this index at Vanguard is through the Tax Managed Small Cap fund.
Vanguard S&P Small-Cap 600 ETF (VIOO)
https://personal.vanguard.com/us/funds/ ... IntExt=INT

User avatar
Doc
Posts: 8752
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: An index isn't an index isn't an index...

Post by Doc » Thu Jan 19, 2017 1:59 pm

Barry Barnitz wrote:A likely candidate for return dispersion vs. other providers:
CRSP also has designed a securities migration approach, called packeting, which lowers turnover and minimizes transaction costs. Packeting allows holdings to be shared between two indexes. A company with a changing market capitalization is moved in separate stages or packets to the next index through a shared market-cap band between them. This helps lower turnover in an index by avoiding the need to move a company all at once when its market cap changes.
Serendipity? There is more than one way to skin the cat front runners.

Vanguard got CRSP to develop a new index methodology to reduce costs just as they got Barclays to "invent" the free float version of their bond indices.

Near the same time that Barry posted I got a Summary Prospectus from Schwab and I even read part of it.
SCHX Schwab US Large-Cap ETF™ Summary Prospectus wrote:The fund will generally give the same weight to a given stock as the index does. However, when the investment adviser believes it is appropriate to do so ... or to address liquidity considerations with respect to a stock, the investment adviser may cause the fund’s weighting of a stock to be more or less than the index’s weighting of the stock. The fund may sell securities that are represented in the index in anticipation of their removal from the index, or buy securities that are not yet represented in the index in anticipation of their addition to the index.
I guess if your mutual funds are large enough you can get the index providers to develop "specialized" indices for you but if you are smaller index fund provider you have to do something else to obtain the same desired effect. :D
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

Post Reply