For now, just an FYI - I received the following Vanguard reply to my enquiry about changes to VDADX's benchmark index
Vanguard has recently selected a new benchmark provider for Vanguard
Dividend Appreciation Index Fund Admiral Shares (VDADX).
VDADX will change its index to the S&P U.S. Dividend Growers Index (S&P
DJI) from the Nasdaq US Dividend Achievers Select Index. The changes will
take effect in the 3rd quarter of 2021.
As part of Vanguard’s ongoing due diligence, we regularly analyze all our
products to ensure that they are delivering on their investment objectives
and giving investors the best chance for investment success. Recently, this
process has highlighted the solid design of the funds, along with some
opportunities for benchmark refinements.
Vanguard has a longstanding partnership with S&P DJI, and employed S&P DJI
benchmarks for 22 index or index-oriented offerings with $1.18 trillion in
assets under management as of March 31, 2020. S&P DJI is a leading provider
of benchmarks and investable indexes designed to help track market
performance, evaluate portfolios, and develop investment strategies.
S&P Dow Jones Indices is a market leader in dividend index strategies, with
more than $91.7 billion in assets under management tracking its dividend
indexes as of March 31, 2021. We believe that S&P DJI’s approach to
dividend growth indexes closely aligns with our view, and that they are
well positioned to administer the indexes for VDADX moving forward.
While we feel that S&P DJI is the optimal partner for VDADX moving forward,
we continue to maintain a significant relationship with Nasdaq, Inc., as a
valuable partner across business lines, including the registration of 24
ETFs encompassing $319.9 billion in assets under management on the Nasdaq
exchange as of March 31, 2021.
Pg 2 -
"The changes to the Funds’ target indexes are not expected to increase Fund
expense ratios. The transition may cause each Fund to realize taxable capital
gains. In the event the transition generates capital gains, the Funds may be
required to distribute capital gains to shareholders.
The Funds are expected to implement the index changes in the third quarter of
2021. The Funds’ prospectuses will be updated at that time to reflect the
changes. To protect the Funds from the potential for harmful “front running” by
traders, the exact timing of the index changes will not be disclosed to investors.
In the meantime, the Funds will continue to seek to track their current indexes."
"The S&P U.S. Dividend Growers Index is designed to measure the performance of U.S. companies that have followed a policy of consistently increasing dividends every year for at least 10 consecutive years. The index excludes the top 25% highest-yielding eligible companies from the index."
The benchmark index used prior to the change to SPGlobal does not appear to have such an exclusion. https://indexes.nasdaqomx.com/docs/Methodology_DVG.pdf
I wondered how the differences in methodologies might effect the composition of the fund as a result of the change. The median market cap appear to go from $169B (Vanguard Portfolio page) to $13B - that is way too drastic change and I wish somebody checks me on this to make sure I have not goofed up. Weight of top 10 components does not change much (goes down from 31.5% to 30.5%) - i.e., well diversified (expect that from Vanguard). Dividend yield does not appear to change much - goes from 1.85% (M* info) to 1.74% - there are probably calc differences to call the 11 basis points a material difference, without digging in further.
VDADX short term and long term total return performances appear not impressive. There seem to be large blend funds from Vanguard that have much better performance and not attracted as much AUM as VDADX ($64B). What is the attraction of VDADX as a large blend fund? Can not possibly be the current dividend yield of 1.75%.
BaluBalu - Sorry, to make a long story short, the median market cap of VDADX is not $169B. It is about $13B. The Vanguard data page is wrong - no surprise.
The reason that M* did not discuss differences between the medians of the two indices is that they are not, in fact, very different, and that (again) the Vanguard data page is wrong,
If you go to the link above, you will see that the mutual fund VDADX holds 247 holdings. Half of 247 is about 123. So if you page forward, 30 stocks at a time, until you see stocks #ing 120 - 150, and take the 123 stock (Whirlpool [WHR] or Hasbro [HAS]; 3rd or 4th from the top, doesn't really matter), you can look up their market caps.
I did look it up. Vanguard (effectively) did not.
Yahoo links above. According to Yahoo, the market cap of WHR is $13.528B, and HAS is $13.496B. Roughly $13.5B.
According to page 3 of the S&P (we-know-how-to-define-median) factsheet, the median market cap of their index is $13.322B, as of 08-31-21.
So, the VDADX median market cap (at $13.5B) is close enough to the S&P new-index median (at $13.3B). Again, the Vanguard median market cap info is wrong. (Vanguard probably doesn't know how to define median.)
If you don't know the definition for median, you might want to apply for a position at Vanguard, since they won't care if you know or not. If you'd like a definition of median, see below.
PS: Makes you wonder about what else is wrong on the Vanguard page, doesn't it?
Vanguard knows how to define median and gives its definition right on its website: https://retirementplans.vanguard.com/VGApp/pe/Glossary?Term=medianmarketcapitalization
Vanguard weights a fund's holdings when calculating its median market cap, just as Vanguard weights a fund's holdings when calculating its mean.
Do we really all know how to define these terms? How about average market cap? For the S&P 500, Fidelity FXAIX reports $539.7B (June 30), M* FXAIX reports $204.8B (July 31), and S&P reports $79.9B (Aug 31). They can't all be right, or can they?
S&P is reporting an unweighted arithmetic mean, Fidelity is reporting a weighted arithmetic mean, and M* is reporting a weighted geometric mean.
S&P 500 fact sheet: navigate to sheet from this page.
Fidelity FXAIX quarterly fund review.
M* portfolio page for FXAIX.
M* definition of average market cap as weighted geometric mean
Getting back to fund median market cap. The Fidelity quarterly report for FXAIX gives both the weighted and unweighted medians: $187.6B and $30.2B respectively. S&P gives "the" median as $32.1B (it's two months more recent than Fidelity's figure). That's a spread not much different from the Vanguard discrepancy.
It's best to be careful when comparing apples and oranges before declaring one of them wrong.
I googled to see in what grade these days they teach concepts or notions of representativeness -
"What grade do they learn mean, median mode?
At the 6-8 level, students can build on what they learned in earlier grades to develop an understanding of concepts such as mean, median, and mode."
BaluBalu - You're welcome, and thank you for agreeing that Vanguard's terminology and definitions are - at best - confusing.
msf - Thank you for confirming that Vanguard's definition of median does not conform to the standard definition of median, and that S&P uses the standard definition of median.
Also - the link to Vanguard's definition that you provided (under retirement plans?, really?) is pretty obscure. If you're going to make up your own terms as Vanguard does, think that Vanguard should label them as such, and do so on the page or section where they are used.
Again ... Vanguard's non-standard use of median makes you wonder what other terms they are fooling around with.
S&P uses what you call the "standard" definition of mean. But it seems to be the only one calculating the S&P 500 average market cap this way. While it reports an average of $79.9B, the S&P 500 index funds reporting give their average market cap as over $500B.
It makes one wonder what other terms the whole fund industry (not just Vanguard) is fooling around with. Or in the alternative, whether S&P is the one out of step in how it calculates a portfolio's average market cap. Because it seems that the standard way of calculating a portfolio's aggregate statistics is to weight its holdings.
A few sample figures for S&P 500 index fund average market caps:
Fidelity FXAIX - $539.7B (June)
Vanguard VFIAX - $543.1B (June)
Franklin SBSPX - $545.9B (July)
iShares WFSPX $608.4B (August)
Schwab goes these funds one better, by giving two very different figures:
$518.7B in its April 30, 2021 semi annual report, where it labels this a weighted average, and
$204.8B on its fund's webpage (July 31, 2021).
Since Schwab doesn't say that this figure is weighted, surely that must mean that this average is unweighted? No, it's weighted also, but it's calculated using a "non-standard" formula.
Is Vanguard really the one that is playing fast and loose with figures?
Median is a little more tricky because you're not really multiplying values, just counting the heavier ones multiple times. Still I suspect there's a similar effect going on here.
The question was asked and answered about Vanguard's and S&P's disclosures in the context of a switch in index providers for its VDADX/VIG vehicles.
Your posts - bringing Fidelity, iShares, and Franklin into the mix, and bringing up, as you say, "average market caps" or mean market caps, rather than the median (which was under discussion) seem to me to be the financial equivalent of "what-aboutism".
Wiki link here: https://en.wikipedia.org/wiki/Whataboutism
Whataboutism or whataboutery (as in "what about…?") is a variant of the tu quoque logical fallacy, which attempts to discredit an opponent's position by charging hypocrisy without directly refuting or disproving the argument.
M-W link here: https://www.merriam-webster.com/words-at-play/whataboutism-origin-meaning
Whataboutism ... is not merely the changing of a subject ("What about the economy?") to deflect away from an earlier subject as a political strategy; it’s essentially a reversal of accusation, arguing that an opponent is guilty of an offense just as egregious or worse than what the original party was accused of doing, however unconnected the offenses may be.
For the life of me, I can't understand why you are responding so vociferously to my original statements or conclusions.
1. There is a standard, well accepted definition of median.
2. Standard (no pun) & Poor's uses this definition in their description of their indices, including the new one that will be adopted by VDADX/VIG.
3. Vanguard does not use this definition.
4. Vanguard uses its own definition of median that is not clearly labelled (as noted by BaluBalu, it should be labelled "Weighted Median"), and Vanguard's (mal)practice can confuse Vanguard investors (such as BaluBalu).
IMHO, if you want to have a discussion about average (rather than median) market cap conventions used by various other market participants - go for it. Perhaps you could start a new thread for that topic.
PS: While I don't want to turn this thread into a discussion of Vanguard's poor website design (good Lord, we have enough of those!) consider the following.
Below is the link to the search results I obtained for 'median', when using the Vanguard website search box. None of them refer to the 'retirement plan' definition that you provided.
(Wonder what the definition for median is when you are not in a retirement plan?)
It would be negligent to suggest that the output for Mean and Median would be similar for a non-homogenous population. If Vanguard wants to report Mean, they should say they are reporting Mean.
No other fund company beats the drum as much as Vanguard about their focus on fund owners and I had taken their word for it. Just because every other fund company may be playing fast and loose with reported numbers does not excuse Vanguard if it were to choose to do so as well, especially for an index fund.
I guess, my expectations of Vanguard need to be dialed down. That is all.
I understand that some investors are emotionally invested in fund companies. My trying to understand VDADX and Vanguard's reporting is not intended to hurt their feelings - please do not take it personally.
Thanks to those that helped me to dig through info and data.
I wondered how the differences in methodologies might effect the composition of the fund as a result of the change.
The methodologies here are of two indexes. Ideally then one would compare the indexes themselves. Unfortunately, since one of those indexes (Nasdaq US Dividend Achievers Select Index) is proprietary, it's impossible to get data directly about the index. (This illustrates why one of the promoted advantages of index funds - transparency via the underling index - is sometimes less than advertised.) So BaluBalu used VDADX as a proxy for the Nasdaq index.
To its credit, Vanguard is moving this fund from an opaque index to a transparent one.
People generally look at funds in terms of average (mean) market cap. When they ask whether a fund is large-, mid-, or small-cap, they are looking at where the fund's average market cap falls. When they do screening on fund market caps, they are screening on average market cap. And so on.
There are two points here. One is that a discussion of average (mean) market caps is on point because the question is about how these two indexes compare. Average market cap is a part of that comparison. The other is that the use of median rather than average (mean) market cap as a basis of comparison seems a little odd given the greater familiarity with mean market cap. For both these reasons, I will continue to discuss mean market cap as well as median.
Vanguard reports the mean market cap of VDADX (as of June 30) to be $262.8B (vs. $262.6B for the underlying index). In comparison, S&P reports the mean market cap of the replacement index to be $53.3B. As this 5:1 ratio is huge, albeit not nearly as large as the 13:1 ratio of median market caps reported, it raises the same question: is this large difference for real?
That question has the same answer for either average: The discrepancy is due to the fact that S&P calculates unweighted averages (both mean and median), while Vanguard calculates weighted averages.
How would one know this? One way would be simply as a result of due diligence - reading the fund's statutory prospectus, hardly something obscure or difficult to find. In it, one sees: As far as knowing that the average (mean) market cap figures given by funds are weighted, I take that as a given. As noted above, most people look at market cap numbers (or market cap ranges/style boxes) when carefully researching funds. So we can assume that most people understand what these figures represent.
Average market caps are a very familiar quantity. Forget about Vanguard. Even Blackrock, another large purveyor of index funds, posts average market cap figures without seeing the need to add a footnote that the averages are weighted. I understand people's zeal in ragging exclusively on Vanguard, but in this case the narrow focus is misplaced.
Because VDADX is a market-cap weighted fund, the proportion of the fund's assets invested in each stock is proportional to the market cap of that stock. So weighting each holding by the dollars invested is tantamount to weighting each holding by its market cap.
A trivial hypothetical can illustrate this: Assume three companies, A (market cap $2B), B (market cap $3B), and C (market cap $4B), and a market cap weighted fund holding these three companies. So their weights would be in the ratio 2:3:4.
We would compute the weighted mean as:
2/9 x $2B (A) + 3/9 x $3B (B) + 4/9 x $4B (C)
Notice that aside from scaling (dividing by 9), the summands are the squares of the market caps: 2², 3² and 4².
Again appealing to the fact that "everyone" understands and works with weighted average market caps, I'll skip illustrating why weighting presents a better picture than relying on unweighted averages.
I hope I've covered all the comments, from whataboutism to why market cap weighting is relevant to the squaring of market caps in computing averages. As well as the stuff in between.
Below is the M* article re the switch.
Note: M* quote page for VIGI has inaccurate info for December 2021 distributions. It overstates dividend (income) distribution by approx the size of the cap gain. Per share total distribution for December quarter per Vanguard is approx $5.4; whereas, M* shows $10.6