Explain VTI Underperformance During its Early Years

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
Topic Author
ptete
Posts: 340
Joined: Mon Nov 22, 2010 3:26 pm

Explain VTI Underperformance During its Early Years

Post by ptete » Tue Aug 09, 2011 6:38 pm

Vanguard VTI and Seligson USA have about 98% correlation between 2006 and 2011*. For some odd reason, VTI underperformed Seligson USA during 2006-2008, the early years. It is a time when VTI did not have much carry-forwarded losses, more somewhere here. Vanguard has systematically improved (not sure whether due to redemption process or some other factor) later.


The blue line contain dividends/capital distributions with VTI (1.6% reinvested, 2% yield with 30% taxes) while the red line contain no dividend with VTI. Seligson USA distributes no dividends or capital gains. The lines are just a differences between VTI and Seligson USA divided by appropriate VTI valuation.

Image


The products are similar: they are both young and created about the same time. Could some explain why there would be such differences between low-cost indexed products in the same sector?

Please, note that Seligson USA has about 0.38% higher management fee.

*does not include ragged data (some small changes possible)
Last edited by ptete on Tue Aug 09, 2011 6:54 pm, edited 2 times in total.

SP-diceman
Posts: 3968
Joined: Sun Oct 05, 2008 9:17 am

Post by SP-diceman » Tue Aug 09, 2011 6:48 pm

The chart didnt print.

User avatar
Topic Author
ptete
Posts: 340
Joined: Mon Nov 22, 2010 3:26 pm

Post by ptete » Tue Aug 09, 2011 6:55 pm

SP-diceman wrote:The chart didnt print.
Thanks, the image was apparently in my cache. I uploaded the image to some image hosting site. It should now work.

User avatar
mas
Posts: 1461
Joined: Tue Feb 20, 2007 12:54 pm

Post by mas » Tue Aug 09, 2011 7:20 pm

Unfortunately the name "Seligson USA" doesn't ring any bells with me. I understand from your other posts that this is probably a non US mutual fund. When I search Morningstar, it comes back with these:
http://quote.morningstar.com/fund/f.asp ... F0000002JH
http://quote.morningstar.com/fund/f.asp ... F0000002N3

Is this what you are referring to, or something different?

It isn't entirely clear to me what the chart is meant to represent, or how to provide insight without knowing any details of Seligson USA.

User avatar
dnaumov
Posts: 495
Joined: Tue Jul 27, 2010 6:04 pm
Location: Finland
Contact:

Post by dnaumov » Tue Aug 09, 2011 8:48 pm

You are comparing apples and oranges, the Seligon North America fund is a large cap North America (meaning both USA and Canada) fund, while VTI is an US Total Market (only USA, large, mid and small) fund.

User avatar
Topic Author
ptete
Posts: 340
Joined: Mon Nov 22, 2010 3:26 pm

Post by ptete » Wed Aug 10, 2011 3:12 am

mas wrote:Unfortunately the name "Seligson USA" doesn't ring any bells with me. I understand from your other posts that this is probably a non US mutual fund. When I search Morningstar, it comes back with these:
http://quote.morningstar.com/fund/f.asp ... F0000002JH
http://quote.morningstar.com/fund/f.asp ... F0000002N3

Is this what you are referring to, or something different?

It isn't entirely clear to me what the chart is meant to represent, or how to provide insight without knowing any details of Seligson USA.
I think I am comparing ACC version of the fund (the data provided apparently by seligson here [1]). Is there a way to compare the seligson products against Vanguard products in Moringstar? I did some comparison here but for some reason it does not find Vanguard products to the comparison.


[1] http://www.bogleheads.org/forum/viewtopic.php?p=1011852
Last edited by ptete on Wed Aug 10, 2011 4:53 am, edited 1 time in total.

User avatar
Topic Author
ptete
Posts: 340
Joined: Mon Nov 22, 2010 3:26 pm

Post by ptete » Wed Aug 10, 2011 4:07 am

dnaumov wrote:You are comparing apples and oranges, the Seligon North America fund is a large cap North America (meaning both USA and Canada) fund, while VTI is an US Total Market (only USA, large, mid and small) fund.
Now, you are exaggerating. You are right about the sector differences but the correlation over the five years is about 98% -- the valuations of the products change very similarly. Did you verify your statement or is this just rhetoric? I have not found yet any proper analysis about the funds by third parties. Does Morningstar.com provide some comparison between the funds?

User avatar
dnaumov
Posts: 495
Joined: Tue Jul 27, 2010 6:04 pm
Location: Finland
Contact:

Post by dnaumov » Wed Aug 10, 2011 6:18 am

ptete wrote:
dnaumov wrote:You are comparing apples and oranges, the Seligon North America fund is a large cap North America (meaning both USA and Canada) fund, while VTI is an US Total Market (only USA, large, mid and small) fund.
Now, you are exaggerating. You are right about the sector differences but the correlation over the five years is about 98% -- the valuations of the products change very similarly. Did you verify your statement or is this just rhetoric? I have not found yet any proper analysis about the funds by third parties. Does Morningstar.com provide some comparison between the funds?
I am not exaggerating, you simply haven't done your research.

Seligson North America list of investments: http://rahastoluotain.seligson.fi/new/i ... sitelang=2
- lists both US and Canadian stocks

Seligson North America on Morningstar: http://www.morningstar.fi/fi/snapshot/s ... 02JH&tab=3
- shows 70,94% megacap, 26,14% largecap. 84,92% USA + 14,90% Canada

I don't understand why you are comparing it to VTI: http://www.morningstar.fi/fi/snapshot/s ... 2DAJ&tab=3
- which is 40,85% megacap, 30,76% largecap, 19,67% midcap, 6,38% smallcap, 2,33% microcap and 100% USA

Additionally, the Seligson fund will have to pay 15% tax on the dividends received to the USA and Canadian tax authorities before those dividends can be internally reinvested back into the fund. In other words, directly comparing these 2 funds is pointless, it's apples and oranges.

User avatar
Topic Author
ptete
Posts: 340
Joined: Mon Nov 22, 2010 3:26 pm

Post by ptete » Wed Aug 10, 2011 6:47 am

dnaumov wrote:I don't understand why you are comparing it to VTI: http://www.morningstar.fi/fi/snapshot/s ... 2DAJ&tab=3
- which is 40,85% megacap, 30,76% largecap, 19,67% midcap, 6,38% smallcap, 2,33% microcap and 100% USA
Because it is useful for investors to know which products to choose. I do know the differences but I also do know that many Seligson products and Vanguard products are very similar according to simple correlation calculations.
dnaumov wrote:Additionally, the Seligson fund will have to pay 15% tax on the dividends received to the USA and Canadian tax authorities before those dividends can be internally reinvested back into the fund.
Yes and private investors have to pay taxes on dividends and capital gains, depending on their domicile. I have used 30% taxation on VTI's distribution which are reinvested back into the funds, according to taxation in Finland. It helps Northern investors with similar tax rules to decide whether they should invest in US products or their local products. The point about "apples and oranges" is not supported by the tax issue, you can see my calculations here [2] and more here [3]. I have scaled it out as the blue and red lines show.

Sorry I do not really understand your point about apples and oranges. Some issues such as SC can be scaled out of the returns if really wanted. Actually, it would be interesting to do to analyze the products deeper. I am not sure whether tax differences such as in redemption process affect the results, why do Vanguards seem to outperform Seligson products in longer term? Can it be explained just by the sectors or some other factor?


[2] http://www.speedyshare.com/files/297988 ... nguard.ods
[3] http://translate.google.com/translate?j ... n-vertailu

User avatar
dnaumov
Posts: 495
Joined: Tue Jul 27, 2010 6:04 pm
Location: Finland
Contact:

Post by dnaumov » Wed Aug 10, 2011 7:07 am

I am merely saying that just because fund A and fund B happen to have a high correlation, it doesn't mean that you should be comparing them directly. I recall reading about asset price movement correlations and it was discovered that one of the assets that tracks the return of S&P500 very very closely was... the price of butter in Mumbai. Does this mean you should potentially consider the option of replacing your S&P500 index funds with Mumbai butter price index fund?

Now obviously, this is an exaggeration, but I hope you get my point.

User avatar
Topic Author
ptete
Posts: 340
Joined: Mon Nov 22, 2010 3:26 pm

Post by ptete » Wed Aug 10, 2011 7:24 am

dnaumov wrote:I am merely saying that just because fund A and fund B happen to have a high correlation, it doesn't mean that you should be comparing them directly. I recall reading about asset price movement correlations and it was discovered that one of the assets that tracks the return of S&P500 very very closely was... the price of butter in Mumbai. Does this mean you should potentially consider the option of replacing your S&P500 index funds with Mumbai butter price index fund?

Now obviously, this is an exaggeration, but I hope you get my point.
Good point. I am investigating a related thing here [1]. I know the correlation premise is a weak point in the discussion so I want to investigate it further. nisiprius actually pointed some points to consider besides just correlation, such as confidence interval. Now I find correlation calculations in Morningstar just media porn. Have to provide better analysis soon...

I will come back to this discussion when I have investigated the premise deeper.


[1] http://www.bogleheads.org/forum/viewtop ... highlight=

User avatar
mas
Posts: 1461
Joined: Tue Feb 20, 2007 12:54 pm

Post by mas » Wed Aug 10, 2011 11:12 pm

ptete wrote:I think I am comparing ACC version of the fund (the data provided apparently by seligson here [1]). Is there a way to compare the seligson products against Vanguard products in Moringstar? I did some comparison here but for some reason it does not find Vanguard products to the comparison.
I was unable to get Morningstar to do the comparison either.

But...
http://www.seligson.fi/resource/english ... l_2009.pdf

Two interesting tidbits:
Seligson & Co North America Index Fund invests in North American
equities tracking the Dow Jones North America Sustainability
Index
.
Please note: Up to July 1, 2008, the Fund was named USA Index
Fund and tracked the Dow Jones Industrial Average index.

HISTORICAL DATA - DJSI North America and DJSI
United States

Fact Sheet


I was able to find all of the relevant symbols on bloomberg.com:
SELUSAI:FH
A1SGI:IND
INDU:IND
VTI:US

User avatar
Topic Author
ptete
Posts: 340
Joined: Mon Nov 22, 2010 3:26 pm

Post by ptete » Thu Aug 11, 2011 10:28 am

mas wrote:I was able to find all of the relevant symbols on bloomberg.com:
SELUSAI:FH
A1SGI:IND
INDU:IND
VTI:US
Be very careful in analyzing international products with different currencies in Bloomberg, they do not adjust the currencies! So investor comparing such products will be strongly mislead. I wrote an article about it here. I did not find any "currency-fix" -button although there are a lot of other bells-and-whistles.


You can see how Bloomberg vizualises the US funds in different currencies:

Image

and below a more correct way to do it with currency adjusting.

Alert! The graph is a bit misleading because I did not correct the ragged data so do not trust the ending part, I am just here showing that the comparison is totally stupid without currency adjusting:

Image

User avatar
dnaumov
Posts: 495
Joined: Tue Jul 27, 2010 6:04 pm
Location: Finland
Contact:

Post by dnaumov » Thu Aug 11, 2011 2:29 pm

ptete wrote: Image
This is actually quite a remarkable chart. I am assuming VTI+DIV is adjusted for the 30% tax?

User avatar
Topic Author
ptete
Posts: 340
Joined: Mon Nov 22, 2010 3:26 pm

Post by ptete » Thu Aug 11, 2011 3:30 pm

dnaumov wrote:
ptete wrote: Image
This is actually quite a remarkable chart. I am assuming VTI+DIV is adjusted for the 30% tax?

The funds are so similar that I see moivre fractals here and there along the parallel lines (yes it is no illusion, it is a natural thing). Well this is what I meant with about 98% correlation, is it amazing?

The small firm Seligson with much smaller sample (about 160 firms?) is strongly competing Vanguard's much larger sample (about 3.6k?)! I found this result striking -- but there is still a lot work to do to really understand the issue deeper...

I used this very rough definition to scale out the dividend:
=arrayformula(G3:G*1.016^(J3:J/365))
,

(which assumes that the dividend after tax is reinvested immediately to the fund in a similar way as Seligson does it. This is not actually totally true in practice among private investors and it would be quite useful to scale out the uncertainty -factor -- to help decide when it is really beneficial to choose S or V?)

where I assumed 2% dividend yield with 30% taxation every year (not totally right but enough). For more accurate results, I would use R because G. Docs does not allow a lot of data (becoming too slow) so have to use shortcuts here and there.

If you want a more accurate description of the impact of dividends, you need to download the exact days and dividend valuations. It would actually be quite interesting task to do next, to know the exact results and compare how the simplifications changed the results from the real ones.

You can review the document here [1], please, note that it may take some time to load things because it fetches a lot of data automatically and process them. (so time-to-time things such as correlation calculations and graphs are wrongly done while it is fetching data)


[1] https://spreadsheets.google.com/spreads ... utput=html

User avatar
dnaumov
Posts: 495
Joined: Tue Jul 27, 2010 6:04 pm
Location: Finland
Contact:

Post by dnaumov » Thu Aug 11, 2011 5:18 pm

ptete wrote:The small firm Seligson with much smaller sample (about 160 firms?) is strongly competing Vanguard's much larger sample (about 3.6k?)! I found this result striking -- but there is still a lot work to do to really understand the issue deeper...
I see it the other way around. The fact that Vanguard is definately keeping up, despite the fact that Seligson fund pays 15% tax on dividends, while a finnish holder of VTI pays 30% tax on dividends is pretty crazy.

User avatar
Topic Author
ptete
Posts: 340
Joined: Mon Nov 22, 2010 3:26 pm

Post by ptete » Thu Aug 11, 2011 11:56 pm

dnaumov wrote:
ptete wrote:The small firm Seligson with much smaller sample (about 160 firms?) is strongly competing Vanguard's much larger sample (about 3.6k?)! I found this result striking -- but there is still a lot work to do to really understand the issue deeper...
I see it the other way around. The fact that Vanguard is definately keeping up, despite the fact that Seligson fund pays 15% tax on dividends, while a finnish holder of VTI pays 30% tax on dividends is pretty crazy.
>>> 0.15*0.02
0.003

about 0.3% extra cost due to dividends to VTI. VTI has 0.36% lower management cost. So the things almost cancel out. Do you mean by crazy that you are worried that Seligson does some sort of evil things such as taking extra risk or something like that?

vwaeo
Posts: 1
Joined: Fri Aug 12, 2011 4:08 am

Post by vwaeo » Fri Aug 12, 2011 6:38 am

If I understand correctly, your original question has two parts. The first is how there can be meaningful performance differences with such high (98%) correlation while the second part asks what is the likely cause of these differences for VTI versus Seligson.

On the correlation question. Do keep in mind that correlation is only a measure of how well two securities TEND to move in the same DIRECTION. It says little about the relative MAGNITUDE of the moves. Even 100% correlation can result in substantial return differences over time. A “perfect” 1.3 times leveraged fund will have 100% correlation to its non-leveraged counterpart, but the return stream will obviously be quite different. While 98% sounds substantially the same as 100%, it actually allows quite a bit of performance deviation to enter into the equation. For example, over the last five years, in the iShares complex alone, there are 18 ETFs that exhibited 98% or higher correlation with VTI. More importantly, the five-year annualized returns for these funds ranged from negative 0.40% (IVE) to positive 5.35% (IWZ and IWF) versus 3.41% for VTI.

Looking into this performance-versus-correlation question within the context of the relative performance differences seen in the two sub-periods (2006-2008 and 2008-2011) does require some constraints to be placed on just how dramatically the performance differential can swing over time. I say this because a large reversal of relative performance between two sub-periods would, by definition drive a lower correlation over the full period. That said, the shift in relative performance in this case was certainly not large enough to degrade the overall correlation.

As for what caused the differences, I think that mas and dnaumov hit it on the head, though I wouldn’t say that you are comparing apples to oranges (maybe fuji apples to granny smith apples, though). Having even a modest allocation to Canada and little to no exposure to mid/small caps more than explains Seligson’s underperformance over the last three years. While we don’t know the composition of the Seligson portfolio during the earlier sub-period (2006-2008), even a modest slant towards either Canada or DJIA (the supposed benchmark prior to 2008) would have provided measurable outperformance versus VTI.

Despite the 98% correlation, these two securities are anything but substantially identical when viewed WITHIN the index realm. They have substantial differences on many fronts (returns, volatility, exposure, currency, etc.) and the fact that Seligson is able to change their benchmark brings up a whole other set of issues regarding passive versus active but that’s a different subject entirely.

User avatar
Taylor Larimore
Advisory Board
Posts: 29551
Joined: Tue Feb 27, 2007 8:09 pm
Location: Miami FL

Using Past Performance ?

Post by Taylor Larimore » Fri Aug 12, 2011 7:08 am

Bogleheads know:
"Buying funds based purely on their past performance is one of the stupidest things an investor can do." -- Jason Zweig, Wall Street Journal columnist.
"Simplicity is the master key to financial success." -- Jack Bogle

User avatar
Topic Author
ptete
Posts: 340
Joined: Mon Nov 22, 2010 3:26 pm

Post by ptete » Fri Aug 12, 2011 8:02 am

vwaeo wrote:Despite the 98% correlation, these two securities are anything but substantially identical when viewed WITHIN the index realm. They have substantial differences on many fronts (returns, volatility, exposure, currency, etc.) and the fact that Seligson is able to change their benchmark brings up a whole other set of issues regarding passive versus active but that’s a different subject entirely.
The last point about index change has bothered me a lot, changing an index to a more fuzzy index is certainly not a thing that I would expect from a passive fund. Do you know whether firms such as Vanguard or iShares have some rules about this kind of style-drifting?

User avatar
Topic Author
ptete
Posts: 340
Joined: Mon Nov 22, 2010 3:26 pm

Post by ptete » Fri Aug 12, 2011 8:10 am

vwaeo wrote:On the correlation question. Do keep in mind that correlation is only a measure of how well two securities TEND to move in the same DIRECTION. It says little about the relative MAGNITUDE of the moves. Even 100% correlation can result in substantial return differences over time. A “perfect” 1.3 times leveraged fund will have 100% correlation to its non-leveraged counterpart, but the return stream will obviously be quite different. While 98% sounds substantially the same as 100%, it actually allows quite a bit of performance deviation to enter into the equation. For example, over the last five years, in the iShares complex alone, there are 18 ETFs that exhibited 98% or higher correlation with VTI. More importantly, the five-year annualized returns for these funds ranged from negative 0.40% (IVE) to positive 5.35% (IWZ and IWF) versus 3.41% for VTI.
Yes, well how can you analyze correlation? You have done a good start in the next move by breaking the interval into different parts but are there more ways to analyze correlation? I have moved the investigation to here [1].
Looking into this performance-versus-correlation question within the context of the relative performance differences seen in the two sub-periods (2006-2008 and 2008-2011) does require some constraints to be placed on just how dramatically the performance differential can swing over time.
Well what about 2009 point? Look how do you separate different "seasons"? It is not trivial, is it just the supremum/maximum and then with certain rule? Such breaking can significantly affect results.

I say this because a large reversal of relative performance between two sub-periods would, by definition drive a lower correlation over the full period. That said, the shift in relative performance in this case was certainly not large enough to degrade the overall correlation.
Can you clarify this? Is the problem about the amount of certain type of data or something else? I am a bit confused where you are aiming at. Do you mean there are two sequences which a bit like cancel one another?

[1] http://www.bogleheads.org/forum/viewtop ... 1313144002

Post Reply