Vanguard New Factor Funds Portfolio Statistics

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
comeinvest
Posts: 170
Joined: Mon Mar 12, 2012 6:57 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by comeinvest » Fri Apr 27, 2018 5:23 pm

jhfenton wrote:
Fri Apr 27, 2018 7:16 am
comeinvest wrote:
Fri Apr 27, 2018 12:55 am
What precisely is a tangible benefit of the new VFVA (Vanguard US Value Factor ETF) vs. the old VBR (small cap value), mid cap value, large cap value, etc.?
The expense ratio of VFVA is about twice that of the old ETFs.
I gather that VFVA does not track an index, but on the other hand the relevant indexes (CRSP US Small Cap Value Index, etc.) are constructed to minimize turnover too, so I'm curious if there is a tangible benefit here.
Also, VFVA seems to follow a strict, rules based approach too. I'm not sure where exactly the "active" part comes into play, or if "active" is just Vanguard's terminology for "non-market cap" or "rules-based factor".
With the old ETFs, the investor has more control over the weighting of large/mid/small depending on individual preferences or overall portfolio.
Finally, I'm not sure if "deeper value" is necessarily a benefit. If deeper value was always better, as sometimes suggested in this forum, one could easily tweak the portfolio construction rules (weighting and cutoff numbers) to achieve arbitrary deep factor exposure, at no additional cost since the data and the computers are already there.
If you actually want "value" exposure, the new VFVA is a more concentrated take on it.

The CRSP value funds have fairly dilute value exposure for two main reasons: (1) They include 50% of the market by "value", so you are getting a lot of stocks with middling value characteristics. (2) They market cap weight the included stocks, potentially diluting the value exposure further.

I have no problem owning the CRSP funds. I own Vanguard Mid-Cap Value Index Admiral in my 401(k), and I have Vanguard Small-Cap Value Index Admiral in two of our retirement accounts. They offer a cheap and almost infinitely scalable mix of beta, value, and, for the mid and small funds, size factors. They have moderate tracking error to the popular indices people see on TV, so they may be easier for folks to hold on to.

Vanguard certainly could tweak the CRSP indexing rules to create "passive" funds with deeper value characteristics. They could limit the indices to 33% of the market by value and weight the holdings by the strength of their value characteristics rather than market cap.

At that point, though, if you simply gave the manager the discretion to determine when to rebalance the index, you'd essentially have the new "active" value factor ETF.

To be honest, if you aren't immediately excited by VFVA upon looking at its construction methodology and its portfolio characteristics, you shouldn't even consider it. It should not be sold to anyone. There is a reason they are marketing them as a tool for advisors.
I think every new investment proposal should receive scrutiny, and investment decisions should never be guided by immediate excitement.

... thanks for confirming the chracteristics of VFVA and the old value funds. I gather that the "deeper value" is the main benefit of VFVA, and that they implement what I understand is some sort of weighting by value (details undisclosed), which seems similar to RAFI's fundamental indexes. However, my questions that naturally arose are not answered yet:

(1) What EXACTLY (if any) is the "active" portion of the algorithm, or the capital allocation? If there is indeed an "active" portion beyond the quantitative algorithm, then why can't the "active" portfolio decisions not be formulated using rules and incorporated into the algo a priori? What would be an example of a scenario where the active manager would intervene and deviate from the rules based algo? How would typical psychological fallacies and biases of active management be avoided when deviating from the rules based algo on an "ad hoc" basis?

(2) What precisely is the justification of the twice as high expense ratio compared to the old value funds?
If it is the deeper value: Value cutoff numbers or weightings are just parameters to a simple computer algorithm, that should carry zero additional intellectual or computational cost. The data set with the corporate financials is already available. Fundamental style indexing (weighted by value, etc.) can be coded into a simple algorithm. Dynamic factor exposure depending on factor valuations can be coded into a simple algorithm and has been done by Research Affiliates.
If it is the additional "active" human management: What is the evidence that this adds value (expected risk-adjusted return) to the investor?

(3) The doctrine "deeper value = better" cannot be universal, because there is a natural upper limit to value tilt (and same for other factors). Drawbacks of extremely deep value could be either reduced diversification, or capacity of the strategy, or others. It would be easy to control the amount of value tilt by adjusting the parameters of the algorithm, at no cost; therefore I'm sure the degree of value tilt was deliberately chosen by the fund managers. I think that those who argue "the more valuey the better" and are constantly measuring and on the lookout for deeper value funds, would have to mention and quantify their optimal degree of value tilt at the same time, for this argument to make sense.

User avatar
matjen
Posts: 1941
Joined: Sun Nov 20, 2011 11:30 pm

Re: "Taylor Swedroe Portfolio"

Post by matjen » Fri Apr 27, 2018 5:42 pm

Taylor Larimore wrote:
Fri Apr 27, 2018 5:17 pm
Matjen:

I am unaware of any "Taylor Swedroe Portfolio." Perhaps you are referring to another "Taylor."

Best wishes.
Taylor
Ha! I could call it the "Larry Larimore Portfolio." The LLP sort of has a ring to it! ;-)
A man is rich in proportion to the number of things he can afford to let alone.

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Fri Apr 27, 2018 7:47 pm

comeinvest wrote:
Fri Apr 27, 2018 5:23 pm
I think every new investment proposal should receive scrutiny, and investment decisions should never be guided by immediate excitement.

... thanks for confirming the chracteristics of VFVA and the old value funds. I gather that the "deeper value" is the main benefit of VFVA, and that they implement what I understand is some sort of weighting by value (details undisclosed), which seems similar to RAFI's fundamental indexes. However, my questions that naturally arose are not answered yet:

(1) What EXACTLY (if any) is the "active" portion of the algorithm, or the capital allocation? If there is indeed an "active" portion beyond the quantitative algorithm, then why can't the "active" portfolio decisions not be formulated using rules and incorporated into the algo a priori? What would be an example of a scenario where the active manager would intervene and deviate from the rules based algo? How would typical psychological fallacies and biases of active management be avoided when deviating from the rules based algo on an "ad hoc" basis?

(2) What precisely is the justification of the twice as high expense ratio compared to the old value funds?
If it is the deeper value: Value cutoff numbers or weightings are just parameters to a simple computer algorithm, that should carry zero additional intellectual or computational cost. The data set with the corporate financials is already available. Fundamental style indexing (weighted by value, etc.) can be coded into a simple algorithm. Dynamic factor exposure depending on factor valuations can be coded into a simple algorithm and has been done by Research Affiliates.
If it is the additional "active" human management: What is the evidence that this adds value (expected risk-adjusted return) to the investor?

(3) The doctrine "deeper value = better" cannot be universal, because there is a natural upper limit to value tilt (and same for other factors). Drawbacks of extremely deep value could be either reduced diversification, or capacity of the strategy, or others. It would be easy to control the amount of value tilt by adjusting the parameters of the algorithm, at no cost; therefore I'm sure the degree of value tilt was deliberately chosen by the fund managers. I think that those who argue "the more valuey the better" and are constantly measuring and on the lookout for deeper value funds, would have to mention and quantify their optimal degree of value tilt at the same time, for this argument to make sense.
I think you knew what I meant by "excited." If the funds' construction doesn't immediately appeal to you, if they don't fit within your investment policy statement, you shouldn't invest in them. I did give VFVA a great deal of scrutiny before investing in it, as evidenced by this thread.

We discussed it earlier in the thread, and I referred to it in my previous comment. The "active" element in these funds is the managerial discretion when to reconstitute the portfolio. They monitor the relevant factor loads daily, but rather than reconstituting the index on a quarterly or semiannual basis, they will rebalance when and only when the portfolio has drifted significantly from the target factor loads. They could change the portfolio in consecutive months if there's a dramatic change in the market. They could go a year without a significant change. They can drop a single stock if its characteristics dramatically change.

No one is denying that Vanguard could have offered very similar funds and labeled them enhanced index funds or smart beta funds or simply passive factor funds. They decided to make them technically active to give the managers a bit of discretion as to timing and to emphasize the "active" nature of the bet you are making investing in these funds.

I don't think anyone ever said that a stronger value load is always better. There are always trade-offs: in tracking error, in sector concentration, in fund capacity, in trading costs, etc. For many folks, though, the CRSP funds are not as value-y as they would like. These funds are an attempt to offer a more pure exposure to value across the market cap spectrum and to expand the offerings to momentum, quality, liquidity, min vol, and a "multifactor" combination of value, momentum, and quality.

Every fund has fixed management and compliance costs. You need extra bodies to manage the fund, produce the prospectus and other regulatory filings, do the accounting, etc. Vanguard's Quantitative Equity group manages several existing funds, but I can imagine they added some additional heads to manage the six new funds. Bottom line: you can't expect a brand new with $7.5 MM in assets to have an expense ratio as low as established funds with billions in assets. If you don't think VFVA is worth the 6 extra basis points compared to VBR and company, then you shouldn't invest in VFVA.

fennewaldaj
Posts: 219
Joined: Sun Oct 22, 2017 11:30 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by fennewaldaj » Fri Apr 27, 2018 8:22 pm

I seem kind of silly to keep emphasizing that it cost twice as much not that it cost 6 basis points more. 6 basis points is not going to be very significant all other things equal.

Theoretical
Posts: 1422
Joined: Tue Aug 19, 2014 10:09 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by Theoretical » Sat Apr 28, 2018 8:59 am

Re: the expense ratio and perspective:

It's .13%, and for $7.5 million in assets, that is $9750 in fees per year. That is remarkably low cost.

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Sat Apr 28, 2018 9:10 am

Theoretical wrote:
Sat Apr 28, 2018 8:59 am
Re: the expense ratio and perspective:

It's .13%, and for $7.5 million in assets, that is $9750 in fees per year. That is remarkably low cost.
That adds an excellent perspective. I should have done the math.

comeinvest
Posts: 170
Joined: Mon Mar 12, 2012 6:57 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by comeinvest » Sun Apr 29, 2018 4:16 am

jhfenton wrote:
Fri Apr 27, 2018 7:47 pm
The "active" element in these funds is the managerial discretion when to reconstitute the portfolio. They monitor the relevant factor loads daily, but rather than reconstituting the index on a quarterly or semiannual basis, they will rebalance when and only when the portfolio has drifted significantly from the target factor loads. They could change the portfolio in consecutive months if there's a dramatic change in the market. They could go a year without a significant change. They can drop a single stock if its characteristics dramatically change.
... I guess this will ever so slightly add to the average factor tilt over time. And yet, I'm not understanding why this cannot be done rules-based. Maybe it is, but the algorithm is not disclosed. Respectfully, I cannot believe somebody is sitting in front of a screen every day all year monitoring valuation changes for a pay of $9750/year and making ad-hoc decisions to delete a security, or rebalance.
Given that most of the alternative indexes have relatively frequent rebalancing periods plus some kind of "buffering" mechanism to avoid turnover for minor changes, I would expect the additional benefit to be minimal.

Despite this aspect which takes away a little bit from transparency, overall I like the funds compared to most competitors, although it is hard to beat the already excellent VBR et. al. in terms of expense ratio, turnover, and past performance (tracking vs. index).

On another note, I'm torn between VFVA (value) and VFMF (multi-factor). VFMF would make sense as the "cost per factor" would be optimized, and turnover per factor minimized; but if one were to believe that "valuations matter", per current RA return estimates https://interactive.researchaffiliates. ... strategies , the other factors seem to drag down the value factor over the next 5 years. Any thoughts?
Last edited by comeinvest on Sun Apr 29, 2018 5:37 am, edited 2 times in total.

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lazyday » Sun Apr 29, 2018 5:28 am

comeinvest wrote:
Sun Apr 29, 2018 4:16 am
Given that most of the alternative indexes have relatively frequent rebalancing periods plus some kind of "buffering" mechanism to avoid turnover for minor changes, I would expect the additional benefit to be minimal.
Research Affiliates estimates US Smallcap index trading costs of .54% a year. (Though just a portion of the new funds are smallcap.)

Vanguard works hard to reduce costs, and probably does better than that. I suppose you could back out ER and revenues from share lending, then compare fund return to index returns as a simple estimate of trading costs.

Increased trading flexibility could reduce those costs further.

Don't forget that the index return itself can be harmed by assets following the index. Front runners might drive prices up before a company enters the index. After the change, index investors might drive the price up higher. The price might stay elevated as long as the company is in the index, reducing dividend yield and earnings yield. Then the reverse as the company leaves the index, and afterwards.

It's been a while since I read RA methodology, those costs might be included in that .54% figure. Of course they are not included if you estimate index tracking error.

comeinvest
Posts: 170
Joined: Mon Mar 12, 2012 6:57 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by comeinvest » Sun Apr 29, 2018 5:54 am

lazyday wrote:
Sun Apr 29, 2018 5:28 am
comeinvest wrote:
Sun Apr 29, 2018 4:16 am
Given that most of the alternative indexes have relatively frequent rebalancing periods plus some kind of "buffering" mechanism to avoid turnover for minor changes, I would expect the additional benefit to be minimal.
Research Affiliates estimates US Smallcap index trading costs of .54% a year. (Though just a portion of the new funds are smallcap.)

Vanguard works hard to reduce costs, and probably does better than that. I suppose you could back out ER and revenues from share lending, then compare fund return to index returns as a simple estimate of trading costs.

Increased trading flexibility could reduce those costs further.

Don't forget that the index return itself can be harmed by assets following the index. Front runners might drive prices up before a company enters the index. After the change, index investors might drive the price up higher. The price might stay elevated as long as the company is in the index, reducing dividend yield and earnings yield. Then the reverse as the company leaves the index, and afterwards.

It's been a while since I read RA methodology, those costs might be included in that .54% figure. Of course they are not included if you estimate index tracking error.
This is all true. I personally am having a hard time estimating actual implementation cost of current ETFs. I read some academic papers that had relatively high figures, assuming usage of optimized trading algorithms, but assuming trades over relatively short time frames (not patient). I hope that the ETFs that I am using for my factor tilts are smart about trading, but I have a hard time coming by that information.
I would think that front-running is more a problem for indexes with large asset bases following that index (S&P 500), vs. proprietary smart beta indexes. Every ETF provider has their own definition of "value", etc., and then there are a gazillion of different combinations of multi-factor indexes, so I would think it would be hard for someone to systematically front-run these based on expected index inclusion or exclusion, or there may not be a profit to make from front-running.
However, even without front-running, an ETF may skew it's own index by lazy trading "around" the index inclusion or exclusion dates, thus partially front-running their own index, and affecting the price of a security when the index accounts for the addition or removal. That would mean "tracking error - ER + securities lending" would underestimate the actual implementation cost, as you say.
A different thing is the possible chasing of the common factors as a whole by an increasing asset base chasing them. It looks like many factor researchers assume the return from the popular factors to be cut in half in the future. However, relative valuations as per Research Affiliates interactive page, and others, suggest otherwise, at least for the popular "value" factor.

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Sun Apr 29, 2018 7:45 am

comeinvest wrote:
Sun Apr 29, 2018 4:16 am
... I guess this will ever so slightly add to the average factor tilt over time. And yet, I'm not understanding why this cannot be done rules-based. Maybe it is, but the algorithm is not disclosed. Respectfully, I cannot believe somebody is sitting in front of a screen every day all year monitoring valuation changes for a pay of $9750/year and making ad-hoc decisions to delete a security, or rebalance.
Given that most of the alternative indexes have relatively frequent rebalancing periods plus some kind of "buffering" mechanism to avoid turnover for minor changes, I would expect the additional benefit to be minimal.
People have had the same debate for years on DFA funds. Are they active or passive? They're technically not index funds, but they're largely formula-driven, with discretion for timing to control costs and factor loads.

I see these ETFs as very close to DFA funds on the active-passive spectrum.
comeinvest wrote:
Sun Apr 29, 2018 4:16 am
Despite this aspect which takes away a little bit from transparency, overall I like the funds compared to most competitors, although it is hard to beat the already excellent VBR et. al. in terms of expense ratio, turnover, and past performance (tracking vs. index).
There are different kinds of transparency. One legal consequence of being active is that Vanguard has to disclose the portfolio holdings daily. It's a requirement enforced by the SEC for all active ETFs. For its index ETFs, Vanguard discloses holdings monthly in arrears.
comeinvest wrote:
Sun Apr 29, 2018 4:16 am
On another note, I'm torn between VFVA (value) and VFMF (multi-factor). VFMF would make sense as the "cost per factor" would be optimized, and turnover per factor minimized; but if one were to believe that "valuations matter", per current RA return estimates https://interactive.researchaffiliates. ... strategies , the other factors seem to drag down the value factor over the next 5 years. Any thoughts?
I'm hedging my bets and using both VFMFX and VFVA. VFVA is likely to be have higher tracking error to the broader market and have larger sector tilts. VFVA also might have higher long-term return, particularly at a time when the difference in valuations between value and growth stocks are at an extreme. Or value might continue to underperform, especially deep value. VFMFX/VFMF is likely to have a return closer to the broader market, but it at least cuts out junky stocks, expensive stocks, and negative-momentum stocks. I also like that it has Admiral Shares. I wish they had Admiral Shares for VFVA.

User avatar
abuss368
Posts: 12842
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: Vanguard New Factor Funds Portfolio Statistics

Post by abuss368 » Sun Apr 29, 2018 9:01 am

matjen wrote:
Fri Apr 27, 2018 7:36 am
jhfenton wrote:
Wed Apr 04, 2018 9:16 am
I agree. VFMFX looks like a great core fund for believers in tilting who also want simplicity. If I didn't have 401(k) constraints, I could see eventually simplifying to a two-fund equity portfolio: VFMFX and VSS (Vanguard FTSE All-World ex-US Small Cap). VSS is an all small- and mid-cap version of Total International, about 80% developed, 20% emerging markets.
I agree. Reminds me of the Taylor Swedroe Portfolio which I champion.
DFA US Vector Equity I DFVEX
DFA World ex US Targeted Val Instl DWUSX (22% or so Emerging Markets)

And the Intermediate Term Bond Fund(s) of your choice.
BND, IEI, MUB, BMBIX, etc.
Do you mean Larry Swedroe?
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lazyday » Sun Apr 29, 2018 9:23 am

comeinvest wrote:
Sun Apr 29, 2018 5:54 am
However, even without front-running, an ETF may skew it's own index by lazy trading "around" the index inclusion or exclusion dates, thus partially front-running their own index, and affecting the price of a security when the index accounts for the addition or removal.
I have wondered if some funds do this.

Another complication: it seems that some ETFs may use creation and redemption baskets to avoid some trading. For example, perhaps a creation basket might contain a large quantity of shares of a company that's entering the index.

I'm not confident on that and don't know the details, but if I remember correctly Rick Ferri confirmed that something along those lines can happen?

And of course Vanguard can trade within its funds, for example if a company leaves the growth index and joins the value index. I believe that some shares can avoid all transaction costs this way.

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lazyday » Sun Apr 29, 2018 9:29 am

jhfenton wrote:
Sun Apr 29, 2018 7:45 am
VFMFX/VFMF is likely to have a return closer to the broader market, but it at least cuts out junky stocks, expensive stocks, and negative-momentum stocks.
But maybe it would, for example, buy a somewhat expensive stock if it had very high momentum and quality. That's what I like about well designed/run multifactor funds, and, maybe more importantly, that they won't buy a slightly cheap stock if it has terrible momentum and quality.

drk
Posts: 756
Joined: Mon Jul 24, 2017 10:33 pm
Location: Seattle

Re: Vanguard New Factor Funds Portfolio Statistics

Post by drk » Sun Apr 29, 2018 11:27 am

abuss368 wrote:
Sun Apr 29, 2018 9:01 am
matjen wrote:
Fri Apr 27, 2018 7:36 am
jhfenton wrote:
Wed Apr 04, 2018 9:16 am
I agree. VFMFX looks like a great core fund for believers in tilting who also want simplicity. If I didn't have 401(k) constraints, I could see eventually simplifying to a two-fund equity portfolio: VFMFX and VSS (Vanguard FTSE All-World ex-US Small Cap). VSS is an all small- and mid-cap version of Total International, about 80% developed, 20% emerging markets.
I agree. Reminds me of the Taylor Swedroe Portfolio which I champion.
DFA US Vector Equity I DFVEX
DFA World ex US Targeted Val Instl DWUSX (22% or so Emerging Markets)

And the Intermediate Term Bond Fund(s) of your choice.
BND, IEI, MUB, BMBIX, etc.
Do you mean Larry Swedroe?
It seems to be a play on words, a portmanteau, a wink towards synthesizing two seemingly contradictory portfolios.

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Sun Apr 29, 2018 11:29 am

lazyday wrote:
Sun Apr 29, 2018 9:23 am
comeinvest wrote:
Sun Apr 29, 2018 5:54 am
However, even without front-running, an ETF may skew it's own index by lazy trading "around" the index inclusion or exclusion dates, thus partially front-running their own index, and affecting the price of a security when the index accounts for the addition or removal.
I have wondered if some funds do this.

Another complication: it seems that some ETFs may use creation and redemption baskets to avoid some trading. For example, perhaps a creation basket might contain a large quantity of shares of a company that's entering the index.

I'm not confident on that and don't know the details, but if I remember correctly Rick Ferri confirmed that something along those lines can happen?

And of course Vanguard can trade within its funds, for example if a company leaves the growth index and joins the value index. I believe that some shares can avoid all transaction costs this way.
ETFs definitely use creation and redemption baskets to handle index changes. Funds even contract with third-parties to create and then redeem baskets specifically to avoid trading inside the fund and creating capital gains. (Three recent articles from Elizabeth Kashner outline how funds manufacture these creations and redemptions. Here is the first.)

You can see from the third article, that there are still trading costs involved, even if the ETF structure offers an extra tool for managing it (and tax avoidance).
lazyday wrote:
Sun Apr 29, 2018 9:29 am
jhfenton wrote:
Sun Apr 29, 2018 7:45 am
VFMFX/VFMF is likely to have a return closer to the broader market, but it at least cuts out junky stocks, expensive stocks, and negative-momentum stocks.
But maybe it would, for example, buy a somewhat expensive stock if it had very high momentum and quality. That's what I like about well designed/run multifactor funds, and, maybe more importantly, that they won't buy a slightly cheap stock if it has terrible momentum and quality.
True. But it will still tend to keep out the extremes. You can end up with some moderately expensive stocks with strong momentum and quality. Intel is currently the second largest holding in VFMFX/VFMF (1.1%) even though it would not be a traditional value stock based on its book value, and it's P/E valuation is roughly in line with the market (trailing P/E 28.52, forward P/E 14.95). I could see Apple appearing in the portfolio at some point when it has positive momentum, because it has reasonable valuations, solid earnings, and a solid balance sheet. (AAPL is the largest holding by far--at 9.5%!--in the iShares sector-neutral value fund VLUE.) But I don't think you would ever see an Amazon appear in the multifactor fund because the valuation extremes are just too high for its composite score to merit inclusion.

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lazyday » Sun Apr 29, 2018 1:23 pm

jhfenton wrote:
Sun Apr 29, 2018 11:29 am
ETFs definitely use creation and redemption baskets to handle index changes. Funds even contract with third-parties to create and then redeem baskets specifically to avoid trading inside the fund and creating capital gains. (Three recent articles from Elizabeth Kashner outline how funds manufacture these creations and redemptions. Here is the first.)

You can see from the third article, that there are still trading costs involved, even if the ETF structure offers an extra tool for managing it (and tax avoidance).
Thanks, I think I had only seen the first article. A bit hard to follow, maybe someday I'll try going through them more slowly. The third has a section "Reverse Engineering the Redemption Basket" which seems to detail a case where some shares leaving the index were disposed of through a customized basket, instead of selling them.

On the multifactor design--of course what I really want is relatively high factor loading for all factors, with low total costs. I'd expect a composite score method to have higher loading than just splitting the portfolio into sections targeting each factor. Which as I recall is what the Goldman Sachs multifactor funds do.

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Sun Apr 29, 2018 1:31 pm

lazyday wrote:
Sun Apr 29, 2018 1:23 pm
On the multifactor design--of course what I really want is relatively high factor loading for all factors, with low total costs. I'd expect a composite score method to have higher loading than just splitting the portfolio into sections targeting each factor. Which as I recall is what the Goldman Sachs multifactor funds do.
I agree. The Goldman Sachs multifactor funds are like kindergarten-level multifactor funds. Each factor is basically a separate fund. You should get better results from a composite approach.

iShares multifactor funds seem to be constructed more intelligently, and they offer them for more asset classes. Vanguard's multifactor fund seems to have the fewest compromises of all, but it's only US at this point.

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lazyday » Sun Apr 29, 2018 1:58 pm

jhfenton wrote:
Sun Apr 29, 2018 1:31 pm
iShares multifactor funds seem to be constructed more intelligently, and they offer them for more asset classes. Vanguard's multifactor fund seems to have the fewest compromises of all, but it's only US at this point.
I suppose the ishares smallcap multifactor funds might be competitive for someone who is ok doing without largecaps entirely, such as for a "Larry Portfolio".

User avatar
matjen
Posts: 1941
Joined: Sun Nov 20, 2011 11:30 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by matjen » Sun Apr 29, 2018 2:06 pm

drk wrote:
Sun Apr 29, 2018 11:27 am
abuss368 wrote:
Sun Apr 29, 2018 9:01 am
matjen wrote:
Fri Apr 27, 2018 7:36 am
jhfenton wrote:
Wed Apr 04, 2018 9:16 am
I agree. VFMFX looks like a great core fund for believers in tilting who also want simplicity. If I didn't have 401(k) constraints, I could see eventually simplifying to a two-fund equity portfolio: VFMFX and VSS (Vanguard FTSE All-World ex-US Small Cap). VSS is an all small- and mid-cap version of Total International, about 80% developed, 20% emerging markets.
I agree. Reminds me of the Taylor Swedroe Portfolio which I champion.
DFA US Vector Equity I DFVEX
DFA World ex US Targeted Val Instl DWUSX (22% or so Emerging Markets)

And the Intermediate Term Bond Fund(s) of your choice.
BND, IEI, MUB, BMBIX, etc.
Do you mean Larry Swedroe?
It seems to be a play on words, a portmanteau, a wink towards synthesizing two seemingly contradictory portfolios.
Exactly! The Taylor Swedroe combines the simplicity of Taylor Larimore’s Three-Fund Portfolio with the small cap value tilt of a Larry Swedroe approved portfolio. Although more expensive than a three-fund, it allows one to have a portfolio that is globally diversified and balanced and tilted without having to worry much about rebalancing within the slice-and-dice elements since those are handled by DFA (or perhaps Vanguard and/or others moving forward). All with three total funds.
A man is rich in proportion to the number of things he can afford to let alone.

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Sun Apr 29, 2018 5:00 pm

lazyday wrote:
Sun Apr 29, 2018 1:58 pm
jhfenton wrote:
Sun Apr 29, 2018 1:31 pm
iShares multifactor funds seem to be constructed more intelligently, and they offer them for more asset classes. Vanguard's multifactor fund seems to have the fewest compromises of all, but it's only US at this point.
I suppose the ishares smallcap multifactor funds might be competitive for someone who is ok doing without largecaps entirely, such as for a "Larry Portfolio".
I periodically look at the international small cap version (ISCF), but I haven't found it compelling. As I said, I do own EMGF, the IShares EDGE MSCI Multifactor Emerging Markets ETF. It has attractive portfolio characteristics.

Image

I set the "Benchmark" in those tables to be Vanguard's VEMAX. On average, EMGF's holdings are a bit cheaper by every forward-looking measure and have better growth metrics. (The style chart ends up looking a bit odd, though. At least it's not what I would expect given the numbers.) Is the portfolio worth 45 bp versus VEMAX at 14 bp? (Or really, since it's in my HSA at TD Ameritrade, it would be SPEM at 11 bp.) I could make a case either way, but I decided it was. (And EMGF is now attracting assets, so its survival is secure.)

EMGF has a short lifespan of just over two years, but If I use the AQR Factors for Global Ex-US on portfoliovisualizer.com, EMGF appears to have impressive value and momentum loads (Rm-Rf 1.16, SMB -0.08, HML 0.73, MOM 0.74). By the same analysis, VEMAX had pretty much no factor loads.

When I look at ISCF at 40 bp compared to VSS at 13 bp, I don't see anything that would justify the added expense and complexity (especially since we don't have another non-Vanguard, non-401(k) account where it would make sense to hold a separate fund). Plus, it is still tiny and struggling to survive.

Running the same portfoliovisualizer.com regression on ISCF likewise shows a negative value loading (-0.33) and only a slight momentum loading (0.19), certainly nothing worth paying up for. VSS is basically neutral on both. ISCF does have smaller stocks in its portfolio (0.68 vs 0.35).

One complication with international multifactor funds is what constraints to put on countries and sectors. How far do you let the algorithm push you away from a "neutral" allocation? It is not hard to end up with 30% of a multifactor fund in South Korea as is the case with the new UEVM (USAA MSCI Emerging Markets Value Momentum Blend Index ETF). (UEVM uses a value and momentum composite for selection and somehow factors volatility into security weighting.)

iShares puts fairly tight constraints on all of its multifactor funds, US and ex-US, to make the funds more palatable to investors. So far, the Vanguard US factor funds are more adventurous, cutting any tie whatsoever to market cap weighting. I don't know how the Quantitative Equity Group would approach international factor investing.

comeinvest
Posts: 170
Joined: Mon Mar 12, 2012 6:57 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by comeinvest » Sun Apr 29, 2018 5:31 pm

jhfenton wrote:
Sun Apr 29, 2018 5:00 pm
...
One complication with international multifactor funds is what constraints to put on countries and sectors. How far do you let the algorithm push you away from a "neutral" allocation? It is not hard to end up with 30% of a multifactor fund in South Korea as is the case with the new UEVM (USAA MSCI Emerging Markets Value Momentum Blend Index ETF). (UEVM uses a value and momentum composite for selection and somehow factors volatility into security weighting.)

iShares puts fairly tight constraints on all of its multifactor funds, US and ex-US, to make the funds more palatable to investors. So far, the Vanguard US factor funds are more adventurous, cutting any tie whatsoever to market cap weighting. I don't know how the Quantitative Equity Group would approach international factor investing.
... I hope they approach it soon. I think we all agree that international and EM factor investing is a disaster with current funds availability, cost, and implementation.

Looking at the UEVM style box at Morningstar, I'm curious how they accomplish the deep value loading with the momentum factor which normally at least neutralizes the nominal valuation tilt. But then again, it is unclear how the style boxes are exactly derived, and which base index they refer to. I find it more and more a futile exercise trying to reverse-engineer factor loads of factor ETFs, particularly multifactor or international. They also change over time for any particular fund.

User avatar
Wade Garrett
Posts: 49
Joined: Wed Jul 10, 2013 2:41 pm
Location: Jasper, MO/Double Deuce

Re: Vanguard New Factor Funds Portfolio Statistics

Post by Wade Garrett » Sun Apr 29, 2018 6:16 pm

jhfenton wrote:
Sun Apr 29, 2018 5:00 pm
iShares puts fairly tight constraints on all of its multifactor funds, US and ex-US, to make the funds more palatable to investors.
Portfolio Visualizer shows the iShares multi factor funds having lower correlations to their total market fund counterparts than do the AQR multi factor funds. I'm puzzled by this. I know correlations can change over time, but I would have assumed that the AQR funds would have stronger factor tilts and be less correlated with the broad market and not vice versa. I view low correlation as positive due to a stronger diversification benefit.

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lazyday » Mon Apr 30, 2018 6:40 am

jhfenton wrote:
Sun Apr 29, 2018 5:00 pm
I set the "Benchmark" in those tables to be Vanguard's VEMAX. On average, EMGF's holdings are a bit cheaper by every forward-looking measure and have better growth metrics.
It does look impressive.

(Briefly, since I may have said this to you before: I’m avoiding EMGF because the smallcap goal probably dilutes the other factor loads. And EM small might be relatively expensive, though data is iffy, and probably more so for large vs giant. I’d be happy to expand on this.)

Compared to VEMAX, Fundamental EM FNDE is much cheaper but doesn’t clearly have worse growth. Compared to EMGF, FNDF is much cheaper, but with much worse growth. And likely worse quality and momentum, though RAFI may help quality a bit. It holds a fair amount of Russia and Brazil, while EMGF holds a lot of China.
EMGF has a short lifespan of just over two years, but If I use the AQR Factors for Global Ex-US on portfoliovisualizer.com,
Using ex-us data on EM funds, are the results at all useful? I get low R-squared and huge confidence intervals. Same for FNDE which is over 4 years old.
When I look at ISCF at 40 bp compared to VSS at 13 bp …. negative value loading (-0.33) and only a slight momentum loading (0.19), certainly nothing worth paying up for
I ran on all four of ISCF’s factors. Only Small (SMB) looks impressive, though alpha may be higher than VSS.

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Mon Apr 30, 2018 8:23 am

lazyday wrote:
Mon Apr 30, 2018 6:40 am
EMGF has a short lifespan of just over two years, but If I use the AQR Factors for Global Ex-US on portfoliovisualizer.com,
Using ex-us data on EM funds, are the results at all useful? I get low R-squared and huge confidence intervals. Same for FNDE which is over 4 years old.
I'm not sure the factor regression data are useful on EM funds. I certainly take them with a grain of salt.

I looked at FNDE when picking an EM ETF for my HSA when I moved it to Lively+TD Ameritrade, but I chose EMGF. The thing that jumped out with FNDE was its large sector weight in energy companies. That wasn't particularly appealing.

JRB22
Posts: 15
Joined: Fri Dec 30, 2016 9:31 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by JRB22 » Mon Apr 30, 2018 10:00 am

How does TLTE compare in your mind? I have been using this to achieve a factor tilt in the EM space, but have yet to be totally satisfied with any of the available options that exclude hiring an advisor.

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lazyday » Mon Apr 30, 2018 1:32 pm

jhfenton, thanks for your opinions above on the EM and Intl Small funds :)

JRB22,
JRB22 wrote:
Mon Apr 30, 2018 10:00 am
How does TLTE compare in your mind?
I just took a very quick look.

Looking at https://www.flexshares.com/funds/TLTE I see in "Apply Factor Tilt Process" that "Weight is reallocated from large cap and growth style boxes to small and value". So it looks like a small value fund, not as many factors targeted as EMGF. I didn't read any detailed methodology though.

ER at .59% is high, but I haven't looked closely at smallcap EM funds to see if you could do better. For example, it might be that you could mix FNDE with a smallcap EM fund to get similar factor exposure with less cost. It might be a pain to figure that out.

lack_ey
Posts: 6613
Joined: Wed Nov 19, 2014 11:55 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lack_ey » Mon Apr 30, 2018 1:49 pm

Wade Garrett wrote:
Sun Apr 29, 2018 6:16 pm
jhfenton wrote:
Sun Apr 29, 2018 5:00 pm
iShares puts fairly tight constraints on all of its multifactor funds, US and ex-US, to make the funds more palatable to investors.
Portfolio Visualizer shows the iShares multi factor funds having lower correlations to their total market fund counterparts than do the AQR multi factor funds. I'm puzzled by this. I know correlations can change over time, but I would have assumed that the AQR funds would have stronger factor tilts and be less correlated with the broad market and not vice versa. I view low correlation as positive due to a stronger diversification benefit.
Are you puzzled by the mechanics or by the strategy AQR is using?

The iShares multifactor funds include size as a factor. They are notably tilted away from market capitalization among stocks selected. The AQR multifactor funds do not.

e.g. with US large (LRGF, QCELX), iShares has top holdings of Accenture, Anthem, Northrop Grumman, Valero Energy, and Phillips. AQR has top holdings of Apple, Microsoft, Amazon, JP Morgan, and Facebook.

That in of itself will create greater active share for the iShares funds by far and reduce correlation with the index, even if not necessarily increasing non-size factor loadings overall.

AQR's approach is a bit by-the-books, basic (which may not be a bad thing)... and relatively expensive.

comeinvest
Posts: 170
Joined: Mon Mar 12, 2012 6:57 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by comeinvest » Mon Apr 30, 2018 1:52 pm

lazyday wrote:
Mon Apr 30, 2018 1:32 pm
jhfenton, thanks for your opinions above on the EM and Intl Small funds :)

JRB22,
JRB22 wrote:
Mon Apr 30, 2018 10:00 am
How does TLTE compare in your mind?
I just took a very quick look.

Looking at https://www.flexshares.com/funds/TLTE I see in "Apply Factor Tilt Process" that "Weight is reallocated from large cap and growth style boxes to small and value". So it looks like a small value fund, not as many factors targeted as EMGF. I didn't read any detailed methodology though.

ER at .59% is high, but I haven't looked closely at smallcap EM funds to see if you could do better. For example, it might be that you could mix FNDE with a smallcap EM fund to get similar factor exposure with less cost. It might be a pain to figure that out.
Interesting how this thread morphed into international factor tilts.
It appears that EMGF at 0.45% ER and more factors is the better deal. However, I'm not 100% sure if more factors, especially momentum, is a good thing, as long as we don't know the implementation cost of the strategies targeted by the ETFs, and their implementation details. EMGF has a relatively low turnover of 36% though, for an ETF that includes momentum.
I'm not happy about the international factor options available. I am currently using EMGF and DGS for part of my EM factor strategy. I am realizing that the soft cost of an international / EM ETF factor strategy is ca. 1.5% p.a.: ca 0.5% expense ratio, 0.5% foreign withholding (non-recoverable in retirement accounts), and an estimated average of 0.5%. trading / implementation cost (varying depending on factor and large vs. small). I'm not happy. With part of my portfolio I am replicating an EM value strategy with individual securities at 0% cost and close to 0% implementation cost.

garlandwhizzer
Posts: 1988
Joined: Fri Aug 06, 2010 3:42 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by garlandwhizzer » Mon Apr 30, 2018 2:32 pm

In effect FNDE is a large/mega cap deep value fund that dramatically overweights severely depressed markets like Russia, Brazil, and to a lesser extent South Korea at present. One assumes that based on fundamental indexes the large cap/mega cap space is cheaper than the SC/MC space at this point in time. According to Morningstar 91% of all its holdings are in the LC/Mega cap space. Also it has 59% value category and only 6% growth. Like most deep value funds, it is very volatile, lots of ups and downs as headlines hit the fortunes of downtrodden Russia, Brazil, and the Energy space where it concentrates. Bottom line--risky and volatile but good long term returns for those with strong stomachs who can hold on through some serious tracking error.

EMGF is interesting: a multi-factor portfolio (Val, Mom, Small, Qual, and Low Vol) its portfolio heavily concentrated (67%) in only 3 east asian countries (China, S. Korea, and Taiwan). Relative to FNDE, EMGF has much lower average cap weight due to small size factor, and holds about equal amounts of growth (due to MOM) and value. EMGF also employs techniques to reduce portfolio turnover and to reduce factor risk (Barra risk factor analysis). It has therefore tended to be significantly less volatile than FNDE and also slightly less than IEMG, the cap weighted MSCI Index. EMGF attempts to take on no more risk than the cap weighted EM index while pursuing a mult-factor approach attempting to outperform. It has not been in existence long enough (about 2 years) to know if it will succeed in this long term.

Both FNDE and EMGF use MSCI Indexes which, unlike Vanguard's EM index, includes S. Korea as EM. Both seriously underweight and overweight some countries and some industries relative to the cap weight MSCI EM index IEMG and do so with a modestly increased cost structure.

Garland Whizzer

User avatar
Wade Garrett
Posts: 49
Joined: Wed Jul 10, 2013 2:41 pm
Location: Jasper, MO/Double Deuce

Re: Vanguard New Factor Funds Portfolio Statistics

Post by Wade Garrett » Mon Apr 30, 2018 3:41 pm

lack_ey wrote:
Mon Apr 30, 2018 1:49 pm
Wade Garrett wrote:
Sun Apr 29, 2018 6:16 pm

Portfolio Visualizer shows the iShares multi factor funds having lower correlations to their total market fund counterparts than do the AQR multi factor funds. I'm puzzled by this.
Are you puzzled by the mechanics or by the strategy AQR is using?

The iShares multifactor funds include size as a factor. They are notably tilted away from market capitalization among stocks selected. The AQR multifactor funds do not.
Puzzled probably not the right word. I just assumed (incorrectly perhaps) that an AQR fund would have stronger momentum and value factor tilts (relative to an iShares fund) and those stronger tilts would result in less correlation to an EM total market fund.

Good point on the size factor being the driver of the lower correlation. I had forgotten that the ishares MF funds all specifically tilt to size while the AQR MF funds do not.

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lazyday » Tue May 01, 2018 3:52 am

comeinvest wrote:
Mon Apr 30, 2018 1:52 pm
It appears that EMGF at 0.45% ER and more factors is the better deal. However, I'm not 100% sure if more factors, especially momentum, is a good thing, as long as we don't know the implementation cost of the strategies targeted by the ETFs, and their implementation details.
More factors may not always be better, but for those of us who want exposure to multiple factors, a well designed and well run multifactor fund is probably the best way to get those exposures, if ER is low enough.

I’m not sure adding mom to a factor fund will always increase implementation costs. Perhaps a fund that only targets value might even see total costs reduced, if a bit of a momentum tilt is added.
With part of my portfolio I am replicating an EM value strategy with individual securities at 0% cost and close to 0% implementation cost.
I’ve done this too. Around 10 years ago I built a multifactor portfolio of global stocks, mostly US but including both Developed and EM. I didn’t bother with Interactive Brokers so only used ADRs, but was concerned that companies with ADRs might trade at a higher price.

I’ve become more lazy over time, and more comfortable with funds like FNDE and FNDF, so in recent years have only sold and not bought individual stocks. If I had the motivation though, I would consider an IB account and my own EM multifactor portfolio.

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lazyday » Tue May 01, 2018 3:56 am

garlandwhizzer wrote:
Mon Apr 30, 2018 2:32 pm
EMGF is interesting: a multi-factor portfolio (Val, Mom, Small, Qual, and Low Vol)
Just a clarification: high quality might be related to lower volatility, but as I recall the fund doesn't directly target low price volatility.

Vashezzo
Posts: 3
Joined: Tue Feb 27, 2018 5:01 am

Re: Vanguard New Factor Funds Portfolio Statistics

Post by Vashezzo » Tue May 01, 2018 6:18 am

Dr. French updated his data library today with the returns data for March. Since these funds only launched in mid-Feb, this data tripled the number of trading days we can use to estimate factor loads. I'm guessing that this information is fairly accurate since their stated goal is to rebalance as needed to maintain their desired factor loads.

I put together a spreadsheet to quickly crunch the numbers, and here's a summary of the results. A link to the spreadsheet is available here . Feel free to double check my math and/or make suggestions/requests.

Code: Select all

Full Sample: 02/14 - 03/29 		
	VFVA	VFMF	VFMO
Factor	Load	Load	Load
Alpha	0.0341	0.0375	0.0035
Mkt-RF	0.9985	0.9416	1.0446
SMB	0.3887	0.3463	0.4388
HML	0.4998	0.5301	0.0146
Mom   	-0.2512	0.1664	0.3353
R-Sq	0.9924	0.9886	0.9915

Code: Select all

 10-Day Rolling Average Loads
	VFVA	VFMF	VFMO
Factor	Load	Load	Load
Alpha	0.0389	0.0266	-0.0108
Mkt-RF	0.9842	0.9377	1.0470
SMB	0.4204	0.3625	0.4643
HML	0.5289	0.5261	-0.0291
Mom   	-0.2559	0.2130	0.3251
R-Sq	0.9892	0.9832	0.9931

Takeaways:
  • All three funds have substantial size loads (.35-.45) (not surprising given the selection methodology, but very nice to see)
  • Momentum Fund has an average momentum load of .30-.35, but generally increased towards an average of .40ish (looking at 10 day rolling periods)
    It also manages to keep a value load close to zero, no going negative - interesting!
  • Value Fund has a consistent value load of .50+, but at the cost of having a negative .25 momentum load
  • Multifactor Fund manages to have a value load nearly identical to the value fund, while keeping average momentum positive (.10-.20 average). The only compromise seems to be that the size load is .35ish instead of the .45ish for the other two.
  • Alphas are all close to zero, so the Carhart model appears to be working quite well
First month and a half seems really promising to me, I'm strongly considering putting some money into one of these.

Theoretical
Posts: 1422
Joined: Tue Aug 19, 2014 10:09 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by Theoretical » Tue May 01, 2018 6:39 am

A 50-50 FNDE/DGS split really neutralizes the extreme sector and country tilts Of FNDE, while remaining heavily Value.

garlandwhizzer
Posts: 1988
Joined: Fri Aug 06, 2010 3:42 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by garlandwhizzer » Tue May 01, 2018 11:38 am

lazyday wrote:

Just a clarification: high quality might be related to lower volatility, but as I recall the fund doesn't directly target low price volatility.
Thanks for the correction, lazyday. EMGF targets only 4 factors: value, momentum, size, and quality.

Garland Whizzer

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Tue May 01, 2018 3:06 pm

Looking at the market close, I noticed that VFVA had 10x the previous high daily volume today, with 124K shares trading, mostly in two huge blocks this afternoon of around 82,500 and 36,700 shares. That has to represent some new share creation, considering the fund only had 100,000 shares outstanding.

comeinvest
Posts: 170
Joined: Mon Mar 12, 2012 6:57 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by comeinvest » Tue May 01, 2018 3:32 pm

lazyday wrote:
Tue May 01, 2018 3:52 am

...
With part of my portfolio I am replicating an EM value strategy with individual securities at 0% cost and close to 0% implementation cost.
I’ve done this too. Around 10 years ago I built a multifactor portfolio of global stocks, mostly US but including both Developed and EM. I didn’t bother with Interactive Brokers so only used ADRs, but was concerned that companies with ADRs might trade at a higher price.

I’ve become more lazy over time, and more comfortable with funds like FNDE and FNDF, so in recent years have only sold and not bought individual stocks. If I had the motivation though, I would consider an IB account and my own EM multifactor portfolio.
Great to see that I'm not the only one with that idea. I don't use a lot of ADRs because the selection would be very limited, and some ADRs come with annual fees as high as 10-20% of dividends, somewhat defeating the purpose of an individual securities portfolio. I don't know how you replicated multiple factors, but I don't have the resources and time to aggregate the needed data, especially for doing it over and over for rebalancing. I just focus on the "value" factor, buy-and-hold value stocks, and assume that "value" companies often stay "valuey" for a long time, resulting in a value-tilted portfolio at minimal implementation or other cost. The low turnover of some value ETFs and active value funds seems to support my assumption. I wouldn't attempt a DIY value strategy for the U.S. market as the cost savings would be close to zero, but for international and EM the savings seem to be sizable. I expect up to 1.5% p.a. cost reduction vs. an ETF based international / EM value strategy.

lazyday
Posts: 3297
Joined: Wed Mar 14, 2007 10:27 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lazyday » Wed May 02, 2018 6:06 am

comeinvest wrote:
Tue May 01, 2018 3:32 pm
some ADRs come with annual fees as high as 10-20% of dividends
I tried to stick with sponsored ADRs, which might be why I never saw fees that high.
I don't know how you replicated multiple factors, but I don't have the resources and time to aggregate the needed data, especially for doing it over and over for rebalancing.
I wasn't creating my own personal quant fund! Just some simplistic lazy stockpicking, buying cheap quality stocks, with a mild preference for above average dividend yield, high earnings growth, low beta, and low price volatility. After some years, I added momentum as a consideration.

If I were starting today, I'd probably mostly look at value, momentum, and quality, in that order of importance.

Also I didn't rebalance often. Mostly when making top down decisions like moving from US to the cheaper ex-US. Sometimes I took shortcuts when selling, such as not looking closely at value or quality. Momentum is easy.
I wouldn't attempt a DIY value strategy for the U.S. market as the cost savings would be close to zero
Agreed. When I began, there were no affordable, well run quality funds for any region. If buying a US fund today to hold longterm, I'd just use Vanguard's new multifactor fund, even though I'd prefer less quality and perhaps more value focus.
for international and EM the savings seem to be sizable
Maybe for largecap EM and even some developed, it's easy for you and I to buy a little without moving the market or even paying much spread, and the story is probably different for an ETF. So for largecap the individual investor may have some real advantage.

Smallcap seems more complicated. I can share my feelings and limited experience if asked.

Sorry if this got too far off topic. If it's better to use another thread, I'd be fine with switching to one of the old "individual stock" threads.

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Wed May 02, 2018 8:55 am

jhfenton wrote:
Tue May 01, 2018 3:06 pm
Looking at the market close, I noticed that VFVA had 10x the previous high daily volume today, with 124K shares trading, mostly in two huge blocks this afternoon of around 82,500 and 36,700 shares. That has to represent some new share creation, considering the fund only had 100,000 shares outstanding.
Eyeballing the holdings through yesterday on vanguard.com compared to what M* shows for 3/31/18, it does indeed look like VFVA more than doubled in size yesterday. If I had to estimate without doing the math, 125,000 shares were added to the existing 100,000, putting the fund at $17.25 MM yesterday.

I wonder who moved so much money into the fund.

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Fri May 04, 2018 2:18 pm

jhfenton wrote:
Wed May 02, 2018 8:55 am
Eyeballing the holdings through yesterday on vanguard.com compared to what M* shows for 3/31/18, it does indeed look like VFVA more than doubled in size yesterday. If I had to estimate without doing the math, 125,000 shares were added to the existing 100,000, putting the fund at $17.25 MM yesterday.

I wonder who moved so much money into the fund.
It took a couple of days for the data to show up, but ETF.com now shows VFVA at $19.08 MM in AUM.

All of the funds:

VFMF - $26.82 MM
VFVA - $19.08 MM
VFMO - $11.52 MM
VFMV - $7.77 MM
VFLQ - $7.73 MM
VFQY - $7.71 MM

The top 3 and their order are not surprising.

lack_ey
Posts: 6613
Joined: Wed Nov 19, 2014 11:55 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lack_ey » Fri May 04, 2018 2:42 pm

jhfenton wrote:
Fri May 04, 2018 2:18 pm
The top 3 and their order are not surprising.
You think? Vanguard is actually offering something others don't have in the liquidity factor fund, and minimum/low volatility products are fairly popular generally (and the same group at Vanguard manages their $2 billion global min vol mutual fund). I could see those potentially outstripping the only three with any inflows so far.

comeinvest
Posts: 170
Joined: Mon Mar 12, 2012 6:57 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by comeinvest » Fri May 04, 2018 5:59 pm

lack_ey wrote:
Fri May 04, 2018 2:42 pm
jhfenton wrote:
Fri May 04, 2018 2:18 pm
The top 3 and their order are not surprising.
You think? Vanguard is actually offering something others don't have in the liquidity factor fund, and minimum/low volatility products are fairly popular generally (and the same group at Vanguard manages their $2 billion global min vol mutual fund). I could see those potentially outstripping the only three with any inflows so far.
Would anyone currently seriously invest in, or tilt to, min vol, given the expected return of -3% p.a. for 5 years according to RA's interactive tool? I know an argument can be made to invest for the long run and stay the course, and RA could be wrong - but still?! (The -3% is the expected factor return p.a., on top of the already bad expected market return.)

comeinvest
Posts: 170
Joined: Mon Mar 12, 2012 6:57 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by comeinvest » Fri May 04, 2018 6:14 pm

lack_ey wrote:
Fri May 04, 2018 2:42 pm
jhfenton wrote:
Fri May 04, 2018 2:18 pm
The top 3 and their order are not surprising.
You think? Vanguard is actually offering something others don't have in the liquidity factor fund, and minimum/low volatility products are fairly popular generally (and the same group at Vanguard manages their $2 billion global min vol mutual fund). I could see those potentially outstripping the only three with any inflows so far.
I was not able to find a meaningful explanation or rationale for the low liquidity fund, in terms of expected returns, risk, and correlations. Intuitively it would appear that the concept of using an ETF to target low liquidity might be counter-productive, as ETFs require liquidity, and more so once they start growing. I'm curious to see statistics, but I would expect that an investor can achieve much deeper "low liquidity", at lower implementation cost, by buying low liquidity equities him/herself.

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Fri May 04, 2018 6:50 pm

comeinvest wrote:
Fri May 04, 2018 5:59 pm
lack_ey wrote:
Fri May 04, 2018 2:42 pm
jhfenton wrote:
Fri May 04, 2018 2:18 pm
The top 3 and their order are not surprising.
You think? Vanguard is actually offering something others don't have in the liquidity factor fund, and minimum/low volatility products are fairly popular generally (and the same group at Vanguard manages their $2 billion global min vol mutual fund). I could see those potentially outstripping the only three with any inflows so far.
Would anyone currently seriously invest in, or tilt to, min vol, given the expected return of -3% p.a. for 5 years according to RA's interactive tool? I know an argument can be made to invest for the long run and stay the course, and RA could be wrong - but still?! (The -3% is the expected factor return p.a., on top of the already bad expected market return.)
I was thinking the same way, but that was my personal bias. lack_ey is probably right that the min vol fund will eventually attract assets.

I should have said that so far they are attracting assets in proportion to how attractive they are to me. :beer
comeinvest wrote:
Fri May 04, 2018 5:59 pm
Would anyone currently seriously invest in, or tilt to, min vol, given the expected return of -3% p.a. for 5 years according to RA's interactive tool? I know an argument can be made to invest for the long run and stay the course, and RA could be wrong - but still?! (The -3% is the expected factor return p.a., on top of the already bad expected market return.)
I don't understand the rationale behind VFLQ either. I can understand why there might a liquidity premium in illiquid microcaps. But I'm sceptical that there is much of a premium in what I would call intermediate liquidity stocks, stocks that rank at the bottom in liquidity measures after screening out those with liquidity that is too low.

Theoretical
Posts: 1422
Joined: Tue Aug 19, 2014 10:09 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by Theoretical » Fri May 04, 2018 7:14 pm

If you listen to their audio interviews and discussions, they view the small size factor as a mix of value and relative liquidity. So it's a factor rather than market cap approach to the historical size premium.

comeinvest
Posts: 170
Joined: Mon Mar 12, 2012 6:57 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by comeinvest » Fri May 04, 2018 7:34 pm

Theoretical wrote:
Fri May 04, 2018 7:14 pm
If you listen to their audio interviews and discussions, they view the small size factor as a mix of value and relative liquidity. So it's a factor rather than market cap approach to the historical size premium.
... I gathered that from the short document on their advisor site that I found, but I have trouble comprehending this generic talk. Need more details. If small, value, and liquidity are correlated, fair enough, you don't have to be a genius to know that smaller companies are less liquid, but then why would I invest in a low liquidity ETF rather than size or value or small value. Give me the stats :) Numbers, please!

lack_ey
Posts: 6613
Joined: Wed Nov 19, 2014 11:55 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lack_ey » Fri May 04, 2018 8:10 pm

comeinvest wrote:
Fri May 04, 2018 7:34 pm
Theoretical wrote:
Fri May 04, 2018 7:14 pm
If you listen to their audio interviews and discussions, they view the small size factor as a mix of value and relative liquidity. So it's a factor rather than market cap approach to the historical size premium.
... I gathered that from the short document on their advisor site that I found, but I have trouble comprehending this generic talk. Need more details. If small, value, and liquidity are correlated, fair enough, you don't have to be a genius to know that smaller companies are less liquid, but then why would I invest in a low liquidity ETF rather than size or value or small value. Give me the stats :) Numbers, please!
One relatively recent paper is "Liquidity as an Investment Style" by Ibbotson et al. (2013, with a 2016 update).

Liquidity should be given equal standing to size, value/growth, and momentum as an investment style. Liquidity, as measured by stock turnover, is an economically significant indicator of long run returns. The returns from liquidity are sufficiently different from the other styles, so that it is not merely a substitute.
Image
Returns for quartiles for different styles in US stocks, 1972-2015.
www.zebracapm.com/files/Liquidity%20as% ... Update.pdf

comeinvest
Posts: 170
Joined: Mon Mar 12, 2012 6:57 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by comeinvest » Fri May 04, 2018 8:23 pm

lack_ey wrote:
Fri May 04, 2018 8:10 pm

One relatively recent paper is "Liquidity as an Investment Style" by Ibbotson et al. (2013, with a 2016 update).
Liquidity should be given equal standing to size, value/growth, and momentum as an investment style. Liquidity, as measured by stock turnover, is an economically significant indicator of long run returns. The returns from liquidity are sufficiently different from the other styles, so that it is not merely a substitute.
Image
Returns for quartiles for different styles in US stocks, 1972-2015.
www.zebracapm.com/files/Liquidity%20as% ... Update.pdf
Thanks. I would have expected some paper from Vanguard specific to the strategy of their new low liquidity fund.
Based on the Ibbotson table above, the trick would be to avoid Q4 of liquidity. But if the liquidity is absolute turnover, not relative to size, wouldn't that largely overlap / correlate with small size? Assuming that a stock with n times the number of shares and n times the market cap and n times as many owners would trade n times often, all other parameters being equal.
The next question would be how it fits into a portfolio. Since it's a long only fund, available capital would be "wasted" on this strategy, that could not be spent on alternate factors and strategies. Vanguard needs to supply facts, correlations, historical and expected returns of what they are doing here.
Last edited by comeinvest on Fri May 04, 2018 8:30 pm, edited 1 time in total.

User avatar
jhfenton
Posts: 3308
Joined: Sat Feb 07, 2015 11:17 am
Location: Ohio

Re: Vanguard New Factor Funds Portfolio Statistics

Post by jhfenton » Fri May 04, 2018 8:29 pm

lack_ey wrote:
Fri May 04, 2018 8:10 pm
Returns for quartiles for different styles in US stocks, 1972-2015.
www.zebracapm.com/files/Liquidity%20as% ... Update.pdf
Thanks for the link to the updated paper. I had found the 2013 journal article after doing some searching. It was linked in several articles on the topic. The debate on liquidity as an independent factor is interesting. I will read more tomorrow. (It's purely out of curiosity. I don't think I'll be buying any VFLQ regardless.)

lack_ey
Posts: 6613
Joined: Wed Nov 19, 2014 11:55 pm

Re: Vanguard New Factor Funds Portfolio Statistics

Post by lack_ey » Fri May 04, 2018 8:45 pm

comeinvest wrote:
Fri May 04, 2018 8:23 pm
Thanks. I would have expected some paper from Vanguard specific to the strategy of their new low liquidity fund.
Based on the Ibbotson table above, the trick would be to avoid Q4 of liquidity. But if the liquidity is absolute turnover, not relative to size, wouldn't that largely overlap / correlate with small size? Assuming that a stock with n times the number of shares and n times the market cap and n times as many owners would trade n times often, all other parameters being equal.
The next question would be how it fits into a portfolio. Since it's a long only fund, available capital would be "wasted" on this strategy, that could not be spent on alternate factors and strategies. Vanguard needs to supply facts, correlations, historical and expected returns of what they are doing here.
Ibbotson measures liquidity by (not absolute) turnover.
Amihud and Mendelson (1986) used bid-ask spreads to explain a cross-section of stock returns. Brennan and Subramanyan (1996) regressed price impact of a unit trade size from microstructure trading data. Amihud (2002) developed a metric using the average price impact relative to daily trading volume of each security. Pástor and Stambaugh (2003) demonstrated that stock returns vary with their sensitivity to marketwide liquidity.

We use stock turnover as our “before the fact” measure of liquidity...

Our methodology consists of a two-part algorithm for the selection (prior) year and the performance (current) year. For each selection year (1971–2014), we examine the top 3,500 U.S. stocks by year-end capitalization. From this universe, we record liquidity as measured by the annual share turnover (the sum of the twelve monthly volumes divided by each month’s shares outstanding)
As mentioned earlier in the thread, Vanguard uses three metrics: percentage turnover, dollar turnover, and Amihud illiquidity.

This construction methodology, with the separate large, mid, and small cap evaluations, results in some weird dispersions in market capitalization of holdings, which I looked at a couple months ago.

Warning: old data. Dot size is fund weighting in a stock. Horizontal axis is forward P/E. Vertical axis is market cap. Left graphic is for Vanguard U.S. Liquidity Factor ETF (VFLQ) and the right for Vanguard Total Stock Market ETF (VTI).

Image Image

(I noticed within the image links, the titles are incorrect; should be forward and not trailing P/E but here we're not interested in the horizontal data anyway)

Post Reply