Four International ETFs that are better diversifiers than VXUS

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Mon Sep 17, 2018 10:56 am

I thought it would be interesting to make a short list of international ETFs that are especially good diversifiers for a balanced portfolio that is otherwise US-centric (e.g. 60/40 Vanguard Total Stock Market ETF & Vanguard Total Bond Market).

After analyzing about 100 ETFs, looking for funds that provided the maximum diversification benefit coupled with a relatively low expense ratio, I came up with these four funds:
  • iShares Edge MSCI Min Vol EAFE ETF (EFAV) ER: 20bps
  • iShares Edge MSCI Intl Size Factor ETF (ISZE) ER: 30bps
  • Vanguard FTSE All-Wld ex-US SmCp ETF (VSS) ER: 13bps
  • Xtrackers MSCI EAFE High Div Yld Eq ETF (HDEF) ER: 20bps
ISZE and VSS are both hold smaller stocks (VSS more than ISZE), whereas EFAV and HDEF both hold larger stocks.

VSS includes some emerging markets stocks, while the other three hold primarily stocks from developed markets.

ISZE and HDEF are the best diversifiers of the the four, and holding both would somewhat improve portfolio diversification over owning either one alone.

VSS has the lowest expense ratio.

With the caveat that some of these funds don't have a long history, I compared to a portfolio of Vanguard Total Stock Market ETF, Vanguard Total Bond Market, & Vanguard Total International Stock ETF to the more diverse portfolio of Vanguard Total Stock Market ETF, Vanguard Total Bond Market, iShares Edge MSCI Intl Size Factor ETF, & Xtrackers MSCI EAFE High Div Yld Eq ETF.

https://www.portfoliovisualizer.com/bac ... tion5_2=30

The more diverse portfolio would have had 21% lower volatility but only 3% lower returns, so much better risk-adjusted returns (Sharpe ratio of 1.70 versus 1.40).

Because large-cap stocks from developed nations have relatively high correlations with US stocks AND tend to dominate the broad international index funds (such as Vanguard Total International Stock ETF), it's my opinion that a US-based investor who wants the benefits of international diversification may be better off selecting their international holdings specifically with an aim of maximizing that diversification benefit. In principle, this would allow the investor to gain the optimal benefit with a lower international allocation. These four funds are all good choices for doing that.

I found it interesting that, at least for the time period I examined, ISZE and HDEF offered better diversification than a broad emerging markets index fund (like VWO or IEMG).
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

ThrustVectoring
Posts: 769
Joined: Wed Jul 12, 2017 2:51 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by ThrustVectoring » Mon Sep 17, 2018 12:23 pm

By default, you want market cap weighting. Anything else is a directional bet on small-cap, low-vol, high dividend, or whatever other slice you're betting on. The point of adding VXUS is that there's no a priori reason to prefer US based large multinationals like Ford over foreign based large multinationals like Toyota.

On a technical note, you seem to be conflating low correlation with diversification. These aren't the same thing. A series of fair coin-flip bets would have zero correlation with VTSAX, but also be of absolutely zero help to your portfolio.
Current portfolio: 60% VTI / 40% VXUS

User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Mon Sep 17, 2018 2:17 pm

ThrustVectoring wrote:
Mon Sep 17, 2018 12:23 pm
By default, you want market cap weighting.
That might be your default. It’s not everyone’s.
ThrustVectoring wrote:
Mon Sep 17, 2018 12:23 pm
On a technical note, you seem to be conflating low correlation with diversification.
Nope. I’m not conflating anything.

Picking four funds with the lowest correlations with VBIAX would have been much easier, but that’s not what I did.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

MotoTrojan
Posts: 6807
Joined: Wed Feb 01, 2017 8:39 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by MotoTrojan » Mon Sep 17, 2018 2:26 pm

Maybe you are confusing correlation in movement with correlation in returns.

VXUS and VTI tend to move in the same directions but not always the same magnitude, otherwise we wouldn't see all these threads about International's poor return.

User avatar
Vulcan
Posts: 1025
Joined: Sat Apr 05, 2014 11:43 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by Vulcan » Mon Sep 17, 2018 2:40 pm

vineviz wrote:
Mon Sep 17, 2018 2:17 pm
ThrustVectoring wrote:
Mon Sep 17, 2018 12:23 pm
By default, you want market cap weighting.
That might be your default. It’s not everyone’s.
You're right. My default is VT.

Can you explain why it shouldn't be?
If you torture the data long enough, it will confess to anything. ~Ronald Coase

User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Mon Sep 17, 2018 2:40 pm

MotoTrojan wrote:
Mon Sep 17, 2018 2:26 pm
Maybe you are confusing correlation in movement with correlation in returns.
Or maybe I’m not confused at all.

Seriously, do you have a question? I’d be happy to explain anything that you are struggling to understand.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Mon Sep 17, 2018 2:43 pm

Vulcan wrote:
Mon Sep 17, 2018 2:40 pm
You're right. My default is VT.

Can you explain why it shouldn't be?
I suspect if you start a topic with that question you’ll get lots of responses. That’s not the topic of this thread.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

livesoft
Posts: 68560
Joined: Thu Mar 01, 2007 8:00 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by livesoft » Mon Sep 17, 2018 2:55 pm

How do DGS and DLS fit in this scheme of things?
Wiki This signature message sponsored by sscritic: Learn to fish.

staythecourse
Posts: 6993
Joined: Mon Jan 03, 2011 9:40 am

Re: Four International ETFs that are better diversifiers than VXUS

Post by staythecourse » Mon Sep 17, 2018 3:12 pm

I've talked about this on this forum several times. I think i great addition to the 3 fund would be kicking out total international and adding a small value international. If one has total stock market or sp 500 I don't see a big advantage of adding another large cap dominated fund even if it is international. I think a great addition is adding a small value international that would be more indicative of each countries local economic climate AND get some tilting AND lower correlation to TSM or SP500.

Good luck.
"The stock market [fluctuation], therefore, is noise. A giant distraction from the business of investing.” | -Jack Bogle

User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Mon Sep 17, 2018 3:42 pm

livesoft wrote:
Mon Sep 17, 2018 2:55 pm
How do DGS and DLS fit in this scheme of things?
They missed my list because they weren't quite as effective at diversification over the period I examined as ISZE or HDEF, and their expense ratios were much higher.

https://www.portfoliovisualizer.com/bac ... tion7_3=40

They definitely are better diversifiers than VXUS, though.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

livesoft
Posts: 68560
Joined: Thu Mar 01, 2007 8:00 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by livesoft » Mon Sep 17, 2018 3:49 pm

So the PV runs are for just only 3 years which is not something I would bet the farm on.
Wiki This signature message sponsored by sscritic: Learn to fish.

HEDGEFUNDIE
Posts: 3631
Joined: Sun Oct 22, 2017 2:06 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by HEDGEFUNDIE » Mon Sep 17, 2018 3:56 pm

Good stuff, this is pretty much how I approach my own portfolio, heavy on the small cap for international, both for return and for diversification.

Now let's talk about whether people should hold hedged currency ETFs :wink:

User avatar
Doc
Posts: 9240
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Four International ETFs that are better diversifiers than VXUS

Post by Doc » Mon Sep 17, 2018 3:57 pm

vineviz wrote:
Mon Sep 17, 2018 10:56 am
Because large-cap stocks from developed nations have relatively high correlations with US stocks AND tend to dominate the broad international index funds (such as Vanguard Total International Stock ETF), it's my opinion that a US-based investor who wants the benefits of international diversification may be better off selecting their international holdings specifically with an aim of maximizing that diversification benefit. In principle, this would allow the investor to gain the optimal benefit with a lower international allocation. These four funds are all good choices for doing that.
I agree with the large-cap developed/US correlation idea.

Did you look at:

Schwab International Small-Cap Eq ETF™ SCHC ER 0.12
Schwab Fdmtl Intl Sm Co Idx SFILX ER 0.39

Personally I don't like VSS because of the emerging markets component but that's only for "sleep tigh"' reasons and the fact that the component is too small to make any difference in the overall picture.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Mon Sep 17, 2018 4:07 pm

livesoft wrote:
Mon Sep 17, 2018 3:49 pm
So the PV runs are for just only 3 years which is not something I would bet the farm on.
I think that's a fair criticism. Balancing a long-term look (which should give a better view of covariances and volatility) with the rapid profusion of new ETFs (the median inception date of the top 25 funds from my initial screen is June of 2015) was a tough challenge.

That said, the evidence is pretty clear that both covariances and volatility are more stable through time than returns are. Since this analysis doesn't explicitly use returns at all (i.e. I wasn't doing mean-variance optimization) I decided that including a broad universe of funds was preferable.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Mon Sep 17, 2018 4:09 pm

Doc wrote:
Mon Sep 17, 2018 3:57 pm
vineviz wrote:
Mon Sep 17, 2018 10:56 am
Because large-cap stocks from developed nations have relatively high correlations with US stocks AND tend to dominate the broad international index funds (such as Vanguard Total International Stock ETF), it's my opinion that a US-based investor who wants the benefits of international diversification may be better off selecting their international holdings specifically with an aim of maximizing that diversification benefit. In principle, this would allow the investor to gain the optimal benefit with a lower international allocation. These four funds are all good choices for doing that.
I agree with the large-cap developed/US correlation idea.

Did you look at:

Schwab International Small-Cap Eq ETF™ SCHC ER 0.12
Schwab Fdmtl Intl Sm Co Idx SFILX ER 0.39

Personally I don't like VSS because of the emerging markets component but that's only for "sleep tigh"' reasons and the fact that the component is too small to make any difference in the overall picture.
I didn't look at SFILX since I only included ETFs.

SCHC just missed the cut on diversification power, but it was pretty close behind VSS.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

User avatar
nisiprius
Advisory Board
Posts: 39424
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Four International ETFs that are better diversifiers than VXUS

Post by nisiprius » Mon Sep 17, 2018 4:14 pm

1) I understand perfectly that "Four International ETFs that are better diversifiers than VXUS" is verbal shorthand for "Four International ETFs that would have been better diversifiers than VXUS," but it would please me if you would edit the thread title to be strictly accurate.

2) In October 2015, an investor might reasonably have said "I think something other than VXUS might be a better diversifier than VXUS, maybe, I don't know, small-cap or low-vol or high-dividend." However, they would not have been able to read your analysis then. How confident are you that you could have identified these four funds, or ones almost as good, in October, 2015?

In the case of Portfolio 1, you chose two out of four funds, probably the two that gave the best results. You could have chosen two in six different ways, so you are showing us best-out-of-six. It might have been instructive to show us the worst, as well, to see if it was almost as good.

3) I am pretty this will demonstrate the same point you want to make equally well, but I would really like you to take the time to adjust Portfolio 1 so that it has the same standard deviation (or whatever you think is the appropriate measure of risk), and show us the superiority of Portfolio 1 in terms of extra return for the same risk rather than reduced risk for the same return.

John C. Bogle has said "An extra percentage point of long-term return is priceless, and an extra percentage point of short-term standard deviation is meaningless." As stated, that remark doesn't quite make sense, but it does point out the important of deciding how much volatility reduction is worth. That seems like a difficult and subjective decision if neither return nor standard deviation is the same. It becomes much easier if we are comparing the return of two portfolios that have had the same standard deviation.

Show us the money--that is, show us the long-term return after equalizing risk.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

fennewaldaj
Posts: 794
Joined: Sun Oct 22, 2017 11:30 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by fennewaldaj » Mon Sep 17, 2018 4:41 pm

I am all for the overweighting small caps idea and I do both domestically and internationally (my portpolio is in the ~50% large 50% small/mid both places) I just can't get on board with the idea of not owning such an important segment of the equity markets at all. So not owning large cap internationally or domestically is a non starter for me. I get the idea of it just not a way I would be comfortable to invest.

User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Mon Sep 17, 2018 4:42 pm

nisiprius wrote:
Mon Sep 17, 2018 4:14 pm
2) In October 2015, an investor might reasonably have said "I think something other than VXUS might be a better diversifier than VXUS, maybe, I don't know, small-cap or low-vol or high-dividend." However, they would not have been able to read your analysis then. How confident are you that you could have identified these four funds, or ones almost as good, in October, 2015?
Pretty confident. As I mentioned earlier, covariances and volatility are both much more stable over time than returns are so finding funds that WILL BE good diversifiers in the future is much more reliably accomplished than finding funds that WILL improve returns (either absolute or risk-adjusted).
nisiprius wrote:
Mon Sep 17, 2018 4:14 pm
In the case of Portfolio 1, you chose two out of four funds, probably the two that gave the best results. You could have chosen two in six different ways, so you are showing us best-out-of-six. It might have been instructive to show us the worst, as well, to see if it was almost as good.
Perhaps. You could definitely combine any of the four funds to produce a portfolio more diversified than just using VXUS, and that's really the thing I'm most interested in illustrating.
nisiprius wrote:
Mon Sep 17, 2018 4:14 pm
I am pretty this will demonstrate the same point you want to make equally well, but I would really like you to take the time to adjust Portfolio 1 so that it has the same standard deviation (or whatever you think is the appropriate measure of risk), and show us the superiority of Portfolio 1 in terms of extra return for the same risk rather than reduced risk for the same return.
You're right that the results would be directionally similar if you delivered Portfolio 2 to match the volatility of Portfolio 1. Adding in t-bills lowers both risk and return, leaving the Sharpe ratio unchanged (though adding the 20% of short term treasuries necessary to deliver the portfolio results in SLIGHTLY more diversification).

It's important to keep in mind that diversification is purely a process of managing risks. It's different from mean-variance optimization, where the goal is to maximize risk-adjusted return.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

MotoTrojan
Posts: 6807
Joined: Wed Feb 01, 2017 8:39 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by MotoTrojan » Mon Sep 17, 2018 5:19 pm

vineviz wrote:
Mon Sep 17, 2018 2:40 pm
MotoTrojan wrote:
Mon Sep 17, 2018 2:26 pm
Maybe you are confusing correlation in movement with correlation in returns.
Or maybe I’m not confused at all.

Seriously, do you have a question? I’d be happy to explain anything that you are struggling to understand.
Why do you feel that VXUS is too similar (not diversified relative) to VTI when their returns have varied significantly over recent timeframes?

User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Mon Sep 17, 2018 9:08 pm

MotoTrojan wrote:
Mon Sep 17, 2018 5:19 pm
Why do you feel that VXUS is too similar (not diversified relative) to VTI when their returns have varied significantly over recent timeframes?
I said "large-cap stocks from developed nations have relatively high correlations with US stocks", and objectively this is true. VXUS is more highly correlated with VTI than funds which hold smaller stocks (like VSS or ISZE), stocks from emerging markets (like VWO), international value stocks (like HDEF), or other passive subclasses (e.g. EFAV).

Apart from correlation, there are several objective measures of portfolio diversification (including Shannon entropy and the diversification ratio). Portfolios mixing US stocks and bonds with international large cap developed funds (like VXUS) perform more poorly on these measures than portfolios containing funds that with lower correlations.

Many investors prefer market cap weighted approaches to investing, and I can respect that choice. For those investors who embrace a modern approach to portfolio construction, though, I thought I'd highlight the funds that screened best.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

User avatar
whodidntante
Posts: 6662
Joined: Thu Jan 21, 2016 11:11 pm
Location: outside the echo chamber

Re: Four International ETFs that are better diversifiers than VXUS

Post by whodidntante » Mon Sep 17, 2018 9:15 pm

With a whopping 5.4 million in AUM, ISZE seems to be an interesting hobby for Blackrock.

User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Mon Sep 17, 2018 9:29 pm

whodidntante wrote:
Mon Sep 17, 2018 9:15 pm
With a whopping 5.4 million in AUM, ISZE seems to be an interesting hobby for Blackrock.
I expect that the ISZE costs BlackRock virtually nothing to manage, since the same quant managers are using the size factor as part of their global and international multifactor products (which have nearly $1.1 billion in AUM).
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

drk
Posts: 1427
Joined: Mon Jul 24, 2017 10:33 pm
Location: Seattle

Re: Four International ETFs that are better diversifiers than VXUS

Post by drk » Mon Sep 17, 2018 10:28 pm

vineviz wrote:
Mon Sep 17, 2018 4:09 pm
I didn't look at SFILX since I only included ETFs.

SCHC just missed the cut on diversification power, but it was pretty close behind VSS.
How about FNDC instead of SFILX? And did you check ISCF?

ThrustVectoring
Posts: 769
Joined: Wed Jul 12, 2017 2:51 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by ThrustVectoring » Tue Sep 18, 2018 5:33 am

vineviz wrote:
Mon Sep 17, 2018 2:17 pm
Nope. I’m not conflating anything.

Picking four funds with the lowest correlations with VBIAX would have been much easier, but that’s not what I did.
What did you do? I don't know how you're defining diversification or what your methodology is.
Current portfolio: 60% VTI / 40% VXUS

User avatar
nisiprius
Advisory Board
Posts: 39424
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Four International ETFs that are better diversifiers than VXUS

Post by nisiprius » Tue Sep 18, 2018 6:00 am

This is the difference we are looking at, i.e. this is the chart I see when I click on the PortfolioVisualizer link vineviz provided.

Image
Last edited by nisiprius on Tue Sep 18, 2018 6:48 am, edited 1 time in total.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

User avatar
nisiprius
Advisory Board
Posts: 39424
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Four International ETFs that are better diversifiers than VXUS

Post by nisiprius » Tue Sep 18, 2018 6:30 am

vineviz named "Vanguard FTSE All-Wld ex-US SmCp ETF (VSS)," as one of the four that backtested best, in a backward-looking exploration of about 100 ETFs. It had inception in 2010. He chose to look at ISZE and HDEF. That had the side effect of cutting the time period to about three years. By using VSS we can at least look at eight years.

To me, the statement that these four international ETFs "are" better diversifiers suggests something robust and consistent. Since VSS was included in his list as a result of backtesting, I should be able to conduct a similar backtest using VSS instead of ISZE and HDEF. I ought to be able to see the result of that better diversification, even if it is not as pronounced.

Using his allocation of 30% US stocks (VTI), 40% US bonds (BND), and 30% international stocks, we see this. Portfolio 1, blue, is using 30% VSS, one of the "four international ETFs that are[sic] better diversifiers than VXUS". Portfolio 2, red, is using 30% VXUS.

Source

Image

Is vineviz' measure of diversification reduced volatility, or is it increased Sharpe ratio?

In this case, using his "better diversifier" would not have reduced volatility. In fact it would have increased it (but microscopically).

It would have increased the Sharpe ratio, a measure of risk-adjusted return (but microscopically).

I judge that the statement that the four named international ETFs "are" better diversifiers than VXUS is fragile.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

typical.investor
Posts: 1257
Joined: Mon Jun 11, 2018 3:17 am

Re: Four International ETFs that are better diversifiers than VXUS

Post by typical.investor » Tue Sep 18, 2018 6:54 am

vineviz wrote:
Mon Sep 17, 2018 10:56 am
I thought it would be interesting to make a short list of international ETFs that are especially good diversifiers for a balanced portfolio that is otherwise US-centric (e.g. 60/40 Vanguard Total Stock Market ETF & Vanguard Total Bond Market).

ISZE and HDEF are the best diversifiers of the the four, and holding both would somewhat improve portfolio diversification over owning either one alone.

The more diverse portfolio would have had 21% lower volatility but only 3% lower returns, so much better risk-adjusted returns (Sharpe ratio of 1.70 versus 1.40).
ISZE and HDEF don't include emerging markets really (1%). So what effect do we expect from leaving out the most volatile assets? Is that really more diversified though?

You can get better returns, the same standard deviation and include emerging markets by using FNDC (dev international), and DGS (emerging). That lacks international large though ...

Portfolio 1
VTI 30.00%
BND 40.00%
ISZE 20.00%
HDEF 10.00%

Portfolio 2
VTI 30.00%
BND 40.00%
VXUS 30.00%

Portfolio 3
VTI 30.00%
BND 40.00%
FNDC (Schwab Fundamental Intl Sm Co ETF) 20.00%
DGS (WisdomTree Emerging Markets SmCp Div ETF) 10.00%

Portfolio CAGR Stdev
Portfolio 1 8.58% 4.48%
Portfolio 2 8.85% 5.66%
Portfolio 3 9.74% 5.65%

https://www.portfoliovisualizer.com/bac ... tion7_3=10

User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Tue Sep 18, 2018 8:40 am

nisiprius wrote:
Tue Sep 18, 2018 6:30 am
Is vineviz' measure of diversification reduced volatility, or is it increased Sharpe ratio?
Neither of those things are a measure of diversification. Reduced volatility is one likely effect of increased diversification, but its important to realize that diversification may or may not increase the ex-post Sharpe ratio. Its also important to realize that volatility can be reduced and/or Sharpe ratio improved by REDUCING the diversification of a portfolio, so neither measure is useful as a proxy for diversification (though understandably an investor might still care about both of those metrics).

There are two relatively easy ways to measure the diversification of a portfolio, both of which I'm sure we've discussed before.

Diversification ratio is perhaps the easiest, since PortfolioVisualizer calculates it directly for you.

Shannon entropy is a little more complicated calculation, but it can be done in Excel using inputs from principal components analysis (which PortfolioVisualizer also does).

I'm not making a claim that the addition diversification from using one of the four funds I mentioned is overwhelmingly dramatic, but it is definitely both measurable and observable.

Here are summary metrics for the two oldest of the four funds I mentioned (VSS and EFAV) compared to VXUS. Looking at just these two allows for a few extra years of analysis.

Image

On both direct measures of diversification, you can see that both VSS and EFAV represent an improvement over VXUS: diversification ratio and Shannon entropy both increase with the use of VSS over VXUS and with the use of EFAV over VSS.

I included some other portfolio metrics, including returns and volatility, in case you are curious about them. Most notably you can see that US market correlation,annualized volatility, maximum drawdown, and conditional value-at-risk were all considerably lower with the EFAV portfolio than with the VXUS portfolio.

Here's the return graph for the three (30% VTI, 40% BND, 30% intl: VXUS in portfolio 1; VSS in portfolio 2; EFAV in portfolio 3).

Image
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

User avatar
Topic Author
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by vineviz » Tue Sep 18, 2018 9:49 am

typical.investor wrote:
Tue Sep 18, 2018 6:54 am
ISZE and HDEF don't include emerging markets really (1%). So what effect do we expect from leaving out the most volatile assets? Is that really more diversified though?
I was honestly surprised that emerging markets funds didn't rise to the top in this analysis. In fact, neither international fund I use in my own portfolio [ISCF (iShares Edge MSCI Mltfct Intl SmCp ETF) and IEMG (iShares Core MSCI Emerging Markets ETF)] made the cut. Each performed well on diversification measures, just not quite as well as the four I listed in the OP.

Note that this research was geared towards finding a single international fund that maximized diversification. It is certainly possible to increase diversification further by using two international funds, in which case I suspect an emerging markets fund might possibly be one of the two. With dozens of low-cost international ETFs, finding the optimal pair from the thousands of possible pairs would be a challenge however.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

klaus14
Posts: 280
Joined: Sun Nov 25, 2018 7:43 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by klaus14 » Sun Oct 27, 2019 5:42 pm

vineviz wrote:
Tue Sep 18, 2018 9:49 am
typical.investor wrote:
Tue Sep 18, 2018 6:54 am
ISZE and HDEF don't include emerging markets really (1%). So what effect do we expect from leaving out the most volatile assets? Is that really more diversified though?
I was honestly surprised that emerging markets funds didn't rise to the top in this analysis. In fact, neither international fund I use in my own portfolio [ISCF (iShares Edge MSCI Mltfct Intl SmCp ETF) and IEMG (iShares Core MSCI Emerging Markets ETF)] made the cut. Each performed well on diversification measures, just not quite as well as the four I listed in the OP.

Note that this research was geared towards finding a single international fund that maximized diversification. It is certainly possible to increase diversification further by using two international funds, in which case I suspect an emerging markets fund might possibly be one of the two. With dozens of low-cost international ETFs, finding the optimal pair from the thousands of possible pairs would be a challenge however.
I am trying to reason about adding min. vol. to my existing multifactor portfolio. i think holding both EFAV and INTF wouldn't make sense because one has negative value loading vs the other positive. both are large cap. their tilts may cancel each other.
However, i guess EFAV would pair well with ISCF since one is large cap heavy and the other is small cap heavy.
so my idea is to move from intf+iscf to efav+iscf.
What's a good way to think about combining these funds?

User avatar
305pelusa
Posts: 982
Joined: Fri Nov 16, 2018 10:20 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by 305pelusa » Sun Oct 27, 2019 8:59 pm

klaus14 wrote:
Sun Oct 27, 2019 5:42 pm
vineviz wrote:
Tue Sep 18, 2018 9:49 am
typical.investor wrote:
Tue Sep 18, 2018 6:54 am
ISZE and HDEF don't include emerging markets really (1%). So what effect do we expect from leaving out the most volatile assets? Is that really more diversified though?
I was honestly surprised that emerging markets funds didn't rise to the top in this analysis. In fact, neither international fund I use in my own portfolio [ISCF (iShares Edge MSCI Mltfct Intl SmCp ETF) and IEMG (iShares Core MSCI Emerging Markets ETF)] made the cut. Each performed well on diversification measures, just not quite as well as the four I listed in the OP.

Note that this research was geared towards finding a single international fund that maximized diversification. It is certainly possible to increase diversification further by using two international funds, in which case I suspect an emerging markets fund might possibly be one of the two. With dozens of low-cost international ETFs, finding the optimal pair from the thousands of possible pairs would be a challenge however.
I am trying to reason about adding min. vol. to my existing multifactor portfolio. i think holding both EFAV and INTF wouldn't make sense because one has negative value loading vs the other positive. both are large cap. their tilts may cancel each other.
However, i guess EFAV would pair well with ISCF since one is large cap heavy and the other is small cap heavy.
so my idea is to move from intf+iscf to efav+iscf.
What's a good way to think about combining these funds?
Beware of using short periods to determine factor exposures. I assume you used the fund itself (started Nov 2011). This is less than 8 years of data; anything goes with these short periods. Fund factor exposure will vary with the years so using more data can be interesting.

Using the underlying index (EAFE Min Vol) since 1990, some example findings:
-From 2000-2010, the index had an HmL of +0.136
-From 2010-2019, the index had an HmL of -0.244

Over the entire span (1990-2019), the HmL was simply +0.025.
Had EFAV been created in 2000 and you had looked at it in 2010, you would've come to the opposite conclusion: that it has a reasonably good value loading.

As Robert T might say, "obviously no guarantees".

Ferdinand2014
Posts: 835
Joined: Mon Dec 17, 2018 6:49 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by Ferdinand2014 » Sun Oct 27, 2019 9:50 pm

vineviz wrote:
Mon Sep 17, 2018 10:56 am
I thought it would be interesting to make a short list of international ETFs that are especially good diversifiers for a balanced portfolio that is otherwise US-centric (e.g. 60/40 Vanguard Total Stock Market ETF & Vanguard Total Bond Market).

After analyzing about 100 ETFs, looking for funds that provided the maximum diversification benefit coupled with a relatively low expense ratio, I came up with these four funds:
  • iShares Edge MSCI Min Vol EAFE ETF (EFAV) ER: 20bps
  • iShares Edge MSCI Intl Size Factor ETF (ISZE) ER: 30bps
  • Vanguard FTSE All-Wld ex-US SmCp ETF (VSS) ER: 13bps
  • Xtrackers MSCI EAFE High Div Yld Eq ETF (HDEF) ER: 20bps
ISZE and VSS are both hold smaller stocks (VSS more than ISZE), whereas EFAV and HDEF both hold larger stocks.

VSS includes some emerging markets stocks, while the other three hold primarily stocks from developed markets.

ISZE and HDEF are the best diversifiers of the the four, and holding both would somewhat improve portfolio diversification over owning either one alone.

VSS has the lowest expense ratio.

With the caveat that some of these funds don't have a long history, I compared to a portfolio of Vanguard Total Stock Market ETF, Vanguard Total Bond Market, & Vanguard Total International Stock ETF to the more diverse portfolio of Vanguard Total Stock Market ETF, Vanguard Total Bond Market, iShares Edge MSCI Intl Size Factor ETF, & Xtrackers MSCI EAFE High Div Yld Eq ETF.

https://www.portfoliovisualizer.com/bac ... tion5_2=30

The more diverse portfolio would have had 21% lower volatility but only 3% lower returns, so much better risk-adjusted returns (Sharpe ratio of 1.70 versus 1.40).

Because large-cap stocks from developed nations have relatively high correlations with US stocks AND tend to dominate the broad international index funds (such as Vanguard Total International Stock ETF), it's my opinion that a US-based investor who wants the benefits of international diversification may be better off selecting their international holdings specifically with an aim of maximizing that diversification benefit. In principle, this would allow the investor to gain the optimal benefit with a lower international allocation. These four funds are all good choices for doing that.

I found it interesting that, at least for the time period I examined, ISZE and HDEF offered better diversification than a broad emerging markets index fund (like VWO or IEMG).
Thoughts on IGRO (iShares international dividend growth ETF er 0.22) vs HDEF? What is the minimum percent to give meaningful diversification?

Also, given EFAV had a turnover of 23% and 9.58% REIT's, I assume holding in tax deferred would be best?
“You only find out who is swimming naked when the tide goes out.“ — Warren Buffett

klaus14
Posts: 280
Joined: Sun Nov 25, 2018 7:43 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by klaus14 » Mon Oct 28, 2019 1:20 am

305pelusa wrote:
Sun Oct 27, 2019 8:59 pm
klaus14 wrote:
Sun Oct 27, 2019 5:42 pm
vineviz wrote:
Tue Sep 18, 2018 9:49 am
typical.investor wrote:
Tue Sep 18, 2018 6:54 am
ISZE and HDEF don't include emerging markets really (1%). So what effect do we expect from leaving out the most volatile assets? Is that really more diversified though?
I was honestly surprised that emerging markets funds didn't rise to the top in this analysis. In fact, neither international fund I use in my own portfolio [ISCF (iShares Edge MSCI Mltfct Intl SmCp ETF) and IEMG (iShares Core MSCI Emerging Markets ETF)] made the cut. Each performed well on diversification measures, just not quite as well as the four I listed in the OP.

Note that this research was geared towards finding a single international fund that maximized diversification. It is certainly possible to increase diversification further by using two international funds, in which case I suspect an emerging markets fund might possibly be one of the two. With dozens of low-cost international ETFs, finding the optimal pair from the thousands of possible pairs would be a challenge however.
I am trying to reason about adding min. vol. to my existing multifactor portfolio. i think holding both EFAV and INTF wouldn't make sense because one has negative value loading vs the other positive. both are large cap. their tilts may cancel each other.
However, i guess EFAV would pair well with ISCF since one is large cap heavy and the other is small cap heavy.
so my idea is to move from intf+iscf to efav+iscf.
What's a good way to think about combining these funds?
Beware of using short periods to determine factor exposures. I assume you used the fund itself (started Nov 2011). This is less than 8 years of data; anything goes with these short periods. Fund factor exposure will vary with the years so using more data can be interesting.

Using the underlying index (EAFE Min Vol) since 1990, some example findings:
-From 2000-2010, the index had an HmL of +0.136
-From 2010-2019, the index had an HmL of -0.244

Over the entire span (1990-2019), the HmL was simply +0.025.
Had EFAV been created in 2000 and you had looked at it in 2010, you would've come to the opposite conclusion: that it has a reasonably good value loading.

As Robert T might say, "obviously no guarantees".
for EFAV i am taking data from msci directly.
Value reads negative here.

they are the ones constructing the index so these are probably coming from underlying stock data. so years of etf history is not needed.

User avatar
unclescrooge
Posts: 3955
Joined: Thu Jun 07, 2012 7:00 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by unclescrooge » Mon Oct 28, 2019 1:55 am

Vulcan wrote:
Mon Sep 17, 2018 2:40 pm
vineviz wrote:
Mon Sep 17, 2018 2:17 pm
ThrustVectoring wrote:
Mon Sep 17, 2018 12:23 pm
By default, you want market cap weighting.
That might be your default. It’s not everyone’s.
You're right. My default is VT.

Can you explain why it shouldn't be?
Because research has shown that market cap weighting is literally the worst way to invest.

If you create portfolio of 30 stocks chosen at random from the universe of stocks, 96% of such portfolios will out perform a market cap weighted portfolio.

In other words, a monkey throwing darts has a good chance of out performing a market cap weighted portfolio.

YRT70
Posts: 398
Joined: Sat Apr 27, 2019 8:51 am

Re: Four International ETFs that are better diversifiers than VXUS

Post by YRT70 » Mon Oct 28, 2019 4:14 am

vineviz wrote:
Mon Sep 17, 2018 3:42 pm
livesoft wrote:
Mon Sep 17, 2018 2:55 pm
How do DGS and DLS fit in this scheme of things?
They missed my list because they weren't quite as effective at diversification over the period I examined as ISZE or HDEF, and their expense ratios were much higher.

https://www.portfoliovisualizer.com/bac ... tion7_3=40

They definitely are better diversifiers than VXUS, though.
But DLS has higher value and size loadings than ISZE. Wouldn't that make it worth the slightly higher expense ratio?

JBeck
Posts: 147
Joined: Fri Apr 15, 2016 4:54 am

Re: Four International ETFs that are better diversifiers than VXUS

Post by JBeck » Mon Oct 28, 2019 7:23 am

vineviz wrote:
Mon Sep 17, 2018 10:56 am
In principle, this would allow the investor to gain the optimal benefit with a lower international allocation. These four funds are all good choices for doing that.
What is your recommended allocation then?

Ferdinand2014
Posts: 835
Joined: Mon Dec 17, 2018 6:49 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by Ferdinand2014 » Mon Oct 28, 2019 7:41 am

unclescrooge wrote:
Mon Oct 28, 2019 1:55 am
Vulcan wrote:
Mon Sep 17, 2018 2:40 pm
vineviz wrote:
Mon Sep 17, 2018 2:17 pm
ThrustVectoring wrote:
Mon Sep 17, 2018 12:23 pm
By default, you want market cap weighting.
That might be your default. It’s not everyone’s.
You're right. My default is VT.

Can you explain why it shouldn't be?
Because research has shown that market cap weighting is literally the worst way to invest.

If you create portfolio of 30 stocks chosen at random from the universe of stocks, 96% of such portfolios will out perform a market cap weighted portfolio.

In other words, a monkey throwing darts has a good chance of out performing a market cap weighted portfolio.
You may be referring to the Cass Business School City University London Study taking multiple equal weighted random stocks (out of 1000 base stocks) and comparing them to a cap weighted index of the same stocks. Their conclusion was a monkey could pick random stocks and do better then a cap weighted index. This study has been completely and thoroughly debunked for the flawed analysis and use of data. First, they picked a time period that by chance favored small cap stocks, second there equal weighting favored small cap stocks resulting in multiple small cap tilted “random” portfolios in a market period where small cap outperformed. The only weighting that will consistently capture the market is cap weighted. By definition. So no, research has not shown cap weighting is inferior to any other strategy.

“Investment strategies with mathematical-sounding names like “quantitative strategies,” “smart beta,” and “four-factor model”—like other investment strategies that are not quantitative—don’t produce superior investment performance, except by chance.”

“Regression formulas alone, however, are not a theory. They are merely patterns perceived in data—barren of explanation, and possibly accidents of randomness.”

— The 3 Simple Rules of Investing: Why Everything You've Heard about Investing Is Wrong — and What to Do Instead by Michael Edesess, Kwok L. Tsui, et al.
http://a.co/c95XsZJ
Last edited by Ferdinand2014 on Mon Oct 28, 2019 7:52 am, edited 1 time in total.
“You only find out who is swimming naked when the tide goes out.“ — Warren Buffett

User avatar
305pelusa
Posts: 982
Joined: Fri Nov 16, 2018 10:20 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by 305pelusa » Mon Oct 28, 2019 7:51 am

klaus14 wrote:
Mon Oct 28, 2019 1:20 am
305pelusa wrote:
Sun Oct 27, 2019 8:59 pm
klaus14 wrote:
Sun Oct 27, 2019 5:42 pm
vineviz wrote:
Tue Sep 18, 2018 9:49 am
typical.investor wrote:
Tue Sep 18, 2018 6:54 am
ISZE and HDEF don't include emerging markets really (1%). So what effect do we expect from leaving out the most volatile assets? Is that really more diversified though?
I was honestly surprised that emerging markets funds didn't rise to the top in this analysis. In fact, neither international fund I use in my own portfolio [ISCF (iShares Edge MSCI Mltfct Intl SmCp ETF) and IEMG (iShares Core MSCI Emerging Markets ETF)] made the cut. Each performed well on diversification measures, just not quite as well as the four I listed in the OP.

Note that this research was geared towards finding a single international fund that maximized diversification. It is certainly possible to increase diversification further by using two international funds, in which case I suspect an emerging markets fund might possibly be one of the two. With dozens of low-cost international ETFs, finding the optimal pair from the thousands of possible pairs would be a challenge however.
I am trying to reason about adding min. vol. to my existing multifactor portfolio. i think holding both EFAV and INTF wouldn't make sense because one has negative value loading vs the other positive. both are large cap. their tilts may cancel each other.
However, i guess EFAV would pair well with ISCF since one is large cap heavy and the other is small cap heavy.
so my idea is to move from intf+iscf to efav+iscf.
What's a good way to think about combining these funds?
Beware of using short periods to determine factor exposures. I assume you used the fund itself (started Nov 2011). This is less than 8 years of data; anything goes with these short periods. Fund factor exposure will vary with the years so using more data can be interesting.

Using the underlying index (EAFE Min Vol) since 1990, some example findings:
-From 2000-2010, the index had an HmL of +0.136
-From 2010-2019, the index had an HmL of -0.244

Over the entire span (1990-2019), the HmL was simply +0.025.
Had EFAV been created in 2000 and you had looked at it in 2010, you would've come to the opposite conclusion: that it has a reasonably good value loading.

As Robert T might say, "obviously no guarantees".
for EFAV i am taking data from msci directly.
Value reads negative here.

they are the ones constructing the index so these are probably coming from underlying stock data. so years of etf history is not needed.
Quick correction, the correct index is actually this one:
https://www.msci.com/documents/10199/a9 ... 92996f17a2

It looks like they might be generating those factor exposure numbers from the index inception itself (2009) based on my results above but it's not clear from the Fact Sheet. If so, 10 years is probably too short of a period to judge the methodology. In fact, they didn't have that chart before:
https://web.archive.org/web/20171204103 ... 92996f17a2

Perhaps they begin to add it after 10 years of actual index data? I'm speculating of course.

TIAX
Posts: 1267
Joined: Sat Jan 11, 2014 12:19 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by TIAX » Mon Oct 28, 2019 7:55 am

Ferdinand2014 wrote:
Mon Oct 28, 2019 7:41 am
unclescrooge wrote:
Mon Oct 28, 2019 1:55 am
Vulcan wrote:
Mon Sep 17, 2018 2:40 pm
vineviz wrote:
Mon Sep 17, 2018 2:17 pm
ThrustVectoring wrote:
Mon Sep 17, 2018 12:23 pm
By default, you want market cap weighting.
That might be your default. It’s not everyone’s.
You're right. My default is VT.

Can you explain why it shouldn't be?
Because research has shown that market cap weighting is literally the worst way to invest.

If you create portfolio of 30 stocks chosen at random from the universe of stocks, 96% of such portfolios will out perform a market cap weighted portfolio.

In other words, a monkey throwing darts has a good chance of out performing a market cap weighted portfolio.
You may be referring to the Cass Business School City University London Study taking multiple equal weighted random stocks (out of 1000 base stocks) and comparing them to a cap weighted index of the same stocks. Their conclusion was a monkey could pick random stocks and do better then a cap weighted index. This study has been completely and thoroughly debunked for the flawed analysis and use of data. First, they picked a time period that by chance favored small cap stocks, second there equal weighting favored small cap stocks resulting in multiple small cap tilted “random” portfolios in a market period where small cap outperformed. The only weighting that will consistently capture the market is cap weighted. By definition. So no, research has not shown cap weighting is inferior to any other strategy.

“Investment strategies with mathematical-sounding names like “quantitative strategies,” “smart beta,” and “four-factor model”—like other investment strategies that are not quantitative—don’t produce superior investment performance, except by chance.”

“Regression formulas alone, however, are not a theory. They are merely patterns perceived in data—barren of explanation, and possibly accidents of randomness.”

— The 3 Simple Rules of Investing: Why Everything You've Heard about Investing Is Wrong — and What to Do Instead by Michael Edesess, Kwok L. Tsui, et al.
http://a.co/c95XsZJ
Bogleheads discussion here.

Ferdinand2014
Posts: 835
Joined: Mon Dec 17, 2018 6:49 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by Ferdinand2014 » Mon Oct 28, 2019 9:23 am

TIAX wrote:
Mon Oct 28, 2019 7:55 am
Ferdinand2014 wrote:
Mon Oct 28, 2019 7:41 am
unclescrooge wrote:
Mon Oct 28, 2019 1:55 am
Vulcan wrote:
Mon Sep 17, 2018 2:40 pm
vineviz wrote:
Mon Sep 17, 2018 2:17 pm


That might be your default. It’s not everyone’s.
You're right. My default is VT.

Can you explain why it shouldn't be?
Because research has shown that market cap weighting is literally the worst way to invest.

If you create portfolio of 30 stocks chosen at random from the universe of stocks, 96% of such portfolios will out perform a market cap weighted portfolio.

In other words, a monkey throwing darts has a good chance of out performing a market cap weighted portfolio.
You may be referring to the Cass Business School City University London Study taking multiple equal weighted random stocks (out of 1000 base stocks) and comparing them to a cap weighted index of the same stocks. Their conclusion was a monkey could pick random stocks and do better then a cap weighted index. This study has been completely and thoroughly debunked for the flawed analysis and use of data. First, they picked a time period that by chance favored small cap stocks, second there equal weighting favored small cap stocks resulting in multiple small cap tilted “random” portfolios in a market period where small cap outperformed. The only weighting that will consistently capture the market is cap weighted. By definition. So no, research has not shown cap weighting is inferior to any other strategy.

“Investment strategies with mathematical-sounding names like “quantitative strategies,” “smart beta,” and “four-factor model”—like other investment strategies that are not quantitative—don’t produce superior investment performance, except by chance.”

“Regression formulas alone, however, are not a theory. They are merely patterns perceived in data—barren of explanation, and possibly accidents of randomness.”

— The 3 Simple Rules of Investing: Why Everything You've Heard about Investing Is Wrong — and What to Do Instead by Michael Edesess, Kwok L. Tsui, et al.
http://a.co/c95XsZJ
Bogleheads discussion here.
I appear to have reinvented the wheel.
“You only find out who is swimming naked when the tide goes out.“ — Warren Buffett

User avatar
Vulcan
Posts: 1025
Joined: Sat Apr 05, 2014 11:43 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by Vulcan » Mon Oct 28, 2019 11:28 am

Ferdinand2014 wrote:
Mon Oct 28, 2019 7:41 am
“Regression formulas alone, however, are not a theory. They are merely patterns perceived in data—barren of explanation, and possibly accidents of randomness.”
:sharebeer

Also, see signature below: vvvvvvv :-)
If you torture the data long enough, it will confess to anything. ~Ronald Coase

User avatar
Vulcan
Posts: 1025
Joined: Sat Apr 05, 2014 11:43 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by Vulcan » Mon Oct 28, 2019 11:33 am

Ferdinand2014 wrote:
Mon Oct 28, 2019 9:23 am
I appear to have reinvented the wheel.
I will take reinventing the wheel over stepping on the same rake any day of the week.
Image
If you torture the data long enough, it will confess to anything. ~Ronald Coase

klaus14
Posts: 280
Joined: Sun Nov 25, 2018 7:43 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by klaus14 » Mon Oct 28, 2019 12:31 pm

305pelusa wrote:
Mon Oct 28, 2019 7:51 am
Quick correction, the correct index is actually this one:
https://www.msci.com/documents/10199/a9 ... 92996f17a2

It looks like they might be generating those factor exposure numbers from the index inception itself (2009) based on my results above but it's not clear from the Fact Sheet. If so, 10 years is probably too short of a period to judge the methodology. In fact, they didn't have that chart before:
https://web.archive.org/web/20171204103 ... 92996f17a2

Perhaps they begin to add it after 10 years of actual index data? I'm speculating of course.
But do they really need to apply regression? They are the ones defining what value is (i believe it was 1/3 book value + 2/3 PE)
And they know these numbers of the underlying stocks for the fund and for the universe. Can't they just compute factor loads directly, from day one? Seems to me like they need no historical data.

User avatar
unclescrooge
Posts: 3955
Joined: Thu Jun 07, 2012 7:00 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by unclescrooge » Mon Oct 28, 2019 12:50 pm

Ferdinand2014 wrote:
Mon Oct 28, 2019 7:41 am
unclescrooge wrote:
Mon Oct 28, 2019 1:55 am
Vulcan wrote:
Mon Sep 17, 2018 2:40 pm
vineviz wrote:
Mon Sep 17, 2018 2:17 pm
ThrustVectoring wrote:
Mon Sep 17, 2018 12:23 pm
By default, you want market cap weighting.
That might be your default. It’s not everyone’s.
You're right. My default is VT.

Can you explain why it shouldn't be?
Because research has shown that market cap weighting is literally the worst way to invest.

If you create portfolio of 30 stocks chosen at random from the universe of stocks, 96% of such portfolios will out perform a market cap weighted portfolio.

In other words, a monkey throwing darts has a good chance of out performing a market cap weighted portfolio.
You may be referring to the Cass Business School City University London Study taking multiple equal weighted random stocks (out of 1000 base stocks) and comparing them to a cap weighted index of the same stocks. Their conclusion was a monkey could pick random stocks and do better then a cap weighted index. This study has been completely and thoroughly debunked for the flawed analysis and use of data. First, they picked a time period that by chance favored small cap stocks, second there equal weighting favored small cap stocks resulting in multiple small cap tilted “random” portfolios in a market period where small cap outperformed. The only weighting that will consistently capture the market is cap weighted. By definition. So no, research has not shown cap weighting is inferior to any other strategy.

“Investment strategies with mathematical-sounding names like “quantitative strategies,” “smart beta,” and “four-factor model”—like other investment strategies that are not quantitative—don’t produce superior investment performance, except by chance.”

“Regression formulas alone, however, are not a theory. They are merely patterns perceived in data—barren of explanation, and possibly accidents of randomness.”

— The 3 Simple Rules of Investing: Why Everything You've Heard about Investing Is Wrong — and What to Do Instead by Michael Edesess, Kwok L. Tsui, et al.
http://a.co/c95XsZJ
It's actually from the book, Heads I win, tails I win, which I highly recommend. The author sirens most of the book recommending market cap weighted index funds. It's only in the last couple of chapters where he mentions the various studies. I'll dig it up and see he he has a reference to the actual study.

User avatar
305pelusa
Posts: 982
Joined: Fri Nov 16, 2018 10:20 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by 305pelusa » Mon Oct 28, 2019 1:02 pm

klaus14 wrote:
Mon Oct 28, 2019 12:31 pm
305pelusa wrote:
Mon Oct 28, 2019 7:51 am
Quick correction, the correct index is actually this one:
https://www.msci.com/documents/10199/a9 ... 92996f17a2

It looks like they might be generating those factor exposure numbers from the index inception itself (2009) based on my results above but it's not clear from the Fact Sheet. If so, 10 years is probably too short of a period to judge the methodology. In fact, they didn't have that chart before:
https://web.archive.org/web/20171204103 ... 92996f17a2

Perhaps they begin to add it after 10 years of actual index data? I'm speculating of course.
But do they really need to apply regression? They are the ones defining what value is (i believe it was 1/3 book value + 2/3 PE)
And they know these numbers of the underlying stocks for the fund and for the universe. Can't they just compute factor loads directly, from day one? Seems to me like they need no historical data.
AFAIK, this fund isn't applying any value or non-value orientations. Its methodology is purely about volatility. Based on their low volatility methodology, they choose stocks.

If you look at the stocks that methodology chose these past 10 years, and then apply a value test to it (whether that's by calculating it with their own definition you provided, or using FF regression), you find that they tended to be growth stocks.

If you apply the same value test to the stocks this methodology would've chosen from 2000-2010 however, they tended to be value-y stocks. Again, regardless of your definition used.

One might then make a reasonable conclusion that this isn't neither a growth or value fund. It's a min volatility fund whose exposure to other values will change over the time due to chance. You might reach other conclusions instead. I'm just giving you the bigger picture instead of just the last decade of the index.

klaus14
Posts: 280
Joined: Sun Nov 25, 2018 7:43 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by klaus14 » Mon Oct 28, 2019 1:12 pm

305pelusa wrote:
Mon Oct 28, 2019 1:02 pm
klaus14 wrote:
Mon Oct 28, 2019 12:31 pm
305pelusa wrote:
Mon Oct 28, 2019 7:51 am
Quick correction, the correct index is actually this one:
https://www.msci.com/documents/10199/a9 ... 92996f17a2

It looks like they might be generating those factor exposure numbers from the index inception itself (2009) based on my results above but it's not clear from the Fact Sheet. If so, 10 years is probably too short of a period to judge the methodology. In fact, they didn't have that chart before:
https://web.archive.org/web/20171204103 ... 92996f17a2

Perhaps they begin to add it after 10 years of actual index data? I'm speculating of course.
But do they really need to apply regression? They are the ones defining what value is (i believe it was 1/3 book value + 2/3 PE)
And they know these numbers of the underlying stocks for the fund and for the universe. Can't they just compute factor loads directly, from day one? Seems to me like they need no historical data.
AFAIK, this fund isn't applying any value or non-value orientations. Its methodology is purely about volatility. Based on their low volatility methodology, they choose stocks.

If you look at the stocks that methodology chose these past 10 years, and then apply a value test to it (whether that's by calculating it with their own definition you provided, or using FF regression), you find that they tended to be growth stocks.

If you apply the same value test to the stocks this methodology would've chosen from 2000-2010 however, they tended to be value-y stocks. Again, regardless of your definition used.

One might then make a reasonable conclusion that this isn't neither a growth or value fund. It's a min volatility fund whose exposure to other values will change over the time due to chance. You might reach other conclusions instead. I'm just giving you the bigger picture instead of just the last decade of the index.
i understand what you are saying. you apply factor regression based on let's say AQR data or Fema French data. However, in this instance, MSCI is playing the role of AQR. They don't need to estimate the factor loads by trying to fit history to some benchmark. They are the ones defining things, they can just compute the p/b and p/e of the underlying stocks. So if they are saying currently fund has negative value loading, we don't need to question that. I think HmL is just book value, i think MSCI definition makes more sense than just using book value.

But if you are saying that current negative load may change in future because this fund doesn't care about value, then of course i would agree. My question was about if current negative value makes it a bad companion to my other positive value loaded funds. i think the answer is yes. And to mitigate that my solution is to use multifactor funds (ISCF) for small cap space and min vol (EFAV) for large cap. in this way, i am getting both factor loadings. What do you think?

User avatar
305pelusa
Posts: 982
Joined: Fri Nov 16, 2018 10:20 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by 305pelusa » Mon Oct 28, 2019 1:28 pm

klaus14 wrote:
Mon Oct 28, 2019 1:12 pm
305pelusa wrote:
Mon Oct 28, 2019 1:02 pm
klaus14 wrote:
Mon Oct 28, 2019 12:31 pm
305pelusa wrote:
Mon Oct 28, 2019 7:51 am
Quick correction, the correct index is actually this one:
https://www.msci.com/documents/10199/a9 ... 92996f17a2

It looks like they might be generating those factor exposure numbers from the index inception itself (2009) based on my results above but it's not clear from the Fact Sheet. If so, 10 years is probably too short of a period to judge the methodology. In fact, they didn't have that chart before:
https://web.archive.org/web/20171204103 ... 92996f17a2

Perhaps they begin to add it after 10 years of actual index data? I'm speculating of course.
But do they really need to apply regression? They are the ones defining what value is (i believe it was 1/3 book value + 2/3 PE)
And they know these numbers of the underlying stocks for the fund and for the universe. Can't they just compute factor loads directly, from day one? Seems to me like they need no historical data.
AFAIK, this fund isn't applying any value or non-value orientations. Its methodology is purely about volatility. Based on their low volatility methodology, they choose stocks.

If you look at the stocks that methodology chose these past 10 years, and then apply a value test to it (whether that's by calculating it with their own definition you provided, or using FF regression), you find that they tended to be growth stocks.

If you apply the same value test to the stocks this methodology would've chosen from 2000-2010 however, they tended to be value-y stocks. Again, regardless of your definition used.

One might then make a reasonable conclusion that this isn't neither a growth or value fund. It's a min volatility fund whose exposure to other values will change over the time due to chance. You might reach other conclusions instead. I'm just giving you the bigger picture instead of just the last decade of the index.
i understand what you are saying. you apply factor regression based on let's say AQR data or Fema French data. However, in this instance, MSCI is playing the role of AQR. They don't need to estimate the factor loads by trying to fit history to some benchmark. They are the ones defining things, they can just compute the p/b and p/e of the underlying stocks. So if they are saying currently fund has negative value loading, we don't need to question that. I think HmL is just book value, i think MSCI definition makes more sense than just using book value.

But if you are saying that current negative load may change in future because this fund doesn't care about value, then of course i would agree. My question was about if current negative value makes it a bad companion to my other positive value loaded funds. i think the answer is yes. And to mitigate that my solution is to use multifactor funds (ISCF) for small cap space and min vol (EFAV) for large cap. in this way, i am getting both factor loadings.
Yes, we're on the same page.

Personally, I don't think the answer to that question is necessarily "yes". Are you looking to hold these funds long term or you'll just switch around based on the current loadings?

Ex: If in 2 years EFAV becomes large, it will "cancel" out ISCF. So then will you switch back to INTF ?

Obviously it depends on your goals. I assumed you're looking to buy and hold these long term. In which case I would place more weight on their long term factor exposure instead of whatever it happened to be on the day you purchased them.

Also, to be clear, if EFAV is growth-y right now, it will "cancel" out whatever value exposure you have, regardless of the fund (INTF or ISCF) used. The value premium does tend to be larger on smaller stocks but that's a bit of a second order effect. Is that what you're banking on?

klaus14
Posts: 280
Joined: Sun Nov 25, 2018 7:43 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by klaus14 » Mon Oct 28, 2019 2:08 pm

305pelusa wrote:
Mon Oct 28, 2019 1:28 pm
klaus14 wrote:
Mon Oct 28, 2019 1:12 pm
305pelusa wrote:
Mon Oct 28, 2019 1:02 pm
klaus14 wrote:
Mon Oct 28, 2019 12:31 pm
305pelusa wrote:
Mon Oct 28, 2019 7:51 am
Quick correction, the correct index is actually this one:
https://www.msci.com/documents/10199/a9 ... 92996f17a2

It looks like they might be generating those factor exposure numbers from the index inception itself (2009) based on my results above but it's not clear from the Fact Sheet. If so, 10 years is probably too short of a period to judge the methodology. In fact, they didn't have that chart before:
https://web.archive.org/web/20171204103 ... 92996f17a2

Perhaps they begin to add it after 10 years of actual index data? I'm speculating of course.
But do they really need to apply regression? They are the ones defining what value is (i believe it was 1/3 book value + 2/3 PE)
And they know these numbers of the underlying stocks for the fund and for the universe. Can't they just compute factor loads directly, from day one? Seems to me like they need no historical data.
AFAIK, this fund isn't applying any value or non-value orientations. Its methodology is purely about volatility. Based on their low volatility methodology, they choose stocks.

If you look at the stocks that methodology chose these past 10 years, and then apply a value test to it (whether that's by calculating it with their own definition you provided, or using FF regression), you find that they tended to be growth stocks.

If you apply the same value test to the stocks this methodology would've chosen from 2000-2010 however, they tended to be value-y stocks. Again, regardless of your definition used.

One might then make a reasonable conclusion that this isn't neither a growth or value fund. It's a min volatility fund whose exposure to other values will change over the time due to chance. You might reach other conclusions instead. I'm just giving you the bigger picture instead of just the last decade of the index.
i understand what you are saying. you apply factor regression based on let's say AQR data or Fema French data. However, in this instance, MSCI is playing the role of AQR. They don't need to estimate the factor loads by trying to fit history to some benchmark. They are the ones defining things, they can just compute the p/b and p/e of the underlying stocks. So if they are saying currently fund has negative value loading, we don't need to question that. I think HmL is just book value, i think MSCI definition makes more sense than just using book value.

But if you are saying that current negative load may change in future because this fund doesn't care about value, then of course i would agree. My question was about if current negative value makes it a bad companion to my other positive value loaded funds. i think the answer is yes. And to mitigate that my solution is to use multifactor funds (ISCF) for small cap space and min vol (EFAV) for large cap. in this way, i am getting both factor loadings.
Yes, we're on the same page.

Personally, I don't think the answer to that question is necessarily "yes". Are you looking to hold these funds long term or you'll just switch around based on the current loadings?

Ex: If in 2 years EFAV becomes large, it will "cancel" out ISCF. So then will you switch back to INTF ?

Obviously it depends on your goals. I assumed you're looking to buy and hold these long term. In which case I would place more weight on their long term factor exposure instead of whatever it happened to be on the day you purchased them.

Also, to be clear, if EFAV is growth-y right now, it will "cancel" out whatever value exposure you have, regardless of the fund (INTF or ISCF) used. The value premium does tend to be larger on smaller stocks but that's a bit of a second order effect. Is that what you're banking on?
Unfortunately ISCF value load is pretty weak: https://www.msci.com/documents/10199/8f ... 798db32f38

I guess what i am trying to do is:
I like min-variance idea and i want to add it to my portfolio. At the same time i am trying to avoid the situation like EFAV overweights stock X but INTF underweights stock X, so they cancel each other and i am paying extra fees for nothing.
I am inspired by VFMF construction: they divide universe into 3 equal parts (small, mid, large) and do factor optimization within each. In my case, i'll be getting min. vol. premium from large and value from small. Now i see that they may still cancel each other but at least it's not as dumb as the scenario i described :) And other factors will stand (small, quality, momentum, low vol)

Is it possible to have both ? I guess you can create your own fund that does min variance with factor constraints (no neg. value) but i don't want to do that :)

User avatar
305pelusa
Posts: 982
Joined: Fri Nov 16, 2018 10:20 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by 305pelusa » Mon Oct 28, 2019 2:30 pm

klaus14 wrote:
Mon Oct 28, 2019 2:08 pm
305pelusa wrote:
Mon Oct 28, 2019 1:28 pm
klaus14 wrote:
Mon Oct 28, 2019 1:12 pm
305pelusa wrote:
Mon Oct 28, 2019 1:02 pm
klaus14 wrote:
Mon Oct 28, 2019 12:31 pm


But do they really need to apply regression? They are the ones defining what value is (i believe it was 1/3 book value + 2/3 PE)
And they know these numbers of the underlying stocks for the fund and for the universe. Can't they just compute factor loads directly, from day one? Seems to me like they need no historical data.
AFAIK, this fund isn't applying any value or non-value orientations. Its methodology is purely about volatility. Based on their low volatility methodology, they choose stocks.

If you look at the stocks that methodology chose these past 10 years, and then apply a value test to it (whether that's by calculating it with their own definition you provided, or using FF regression), you find that they tended to be growth stocks.

If you apply the same value test to the stocks this methodology would've chosen from 2000-2010 however, they tended to be value-y stocks. Again, regardless of your definition used.

One might then make a reasonable conclusion that this isn't neither a growth or value fund. It's a min volatility fund whose exposure to other values will change over the time due to chance. You might reach other conclusions instead. I'm just giving you the bigger picture instead of just the last decade of the index.
i understand what you are saying. you apply factor regression based on let's say AQR data or Fema French data. However, in this instance, MSCI is playing the role of AQR. They don't need to estimate the factor loads by trying to fit history to some benchmark. They are the ones defining things, they can just compute the p/b and p/e of the underlying stocks. So if they are saying currently fund has negative value loading, we don't need to question that. I think HmL is just book value, i think MSCI definition makes more sense than just using book value.

But if you are saying that current negative load may change in future because this fund doesn't care about value, then of course i would agree. My question was about if current negative value makes it a bad companion to my other positive value loaded funds. i think the answer is yes. And to mitigate that my solution is to use multifactor funds (ISCF) for small cap space and min vol (EFAV) for large cap. in this way, i am getting both factor loadings.
Yes, we're on the same page.

Personally, I don't think the answer to that question is necessarily "yes". Are you looking to hold these funds long term or you'll just switch around based on the current loadings?

Ex: If in 2 years EFAV becomes large, it will "cancel" out ISCF. So then will you switch back to INTF ?

Obviously it depends on your goals. I assumed you're looking to buy and hold these long term. In which case I would place more weight on their long term factor exposure instead of whatever it happened to be on the day you purchased them.

Also, to be clear, if EFAV is growth-y right now, it will "cancel" out whatever value exposure you have, regardless of the fund (INTF or ISCF) used. The value premium does tend to be larger on smaller stocks but that's a bit of a second order effect. Is that what you're banking on?
Unfortunately ISCF value load is pretty weak: https://www.msci.com/documents/10199/8f ... 798db32f38

I guess what i am trying to do is:
I like min-variance idea and i want to add it to my portfolio. At the same time i am trying to avoid the situation like EFAV overweights stock X but INTF underweights stock X, so they cancel each other and i am paying extra fees for nothing.
I am inspired by VFMF construction: they divide universe into 3 equal parts (small, mid, large) and do factor optimization within each. In my case, i'll be getting min. vol. premium from large and value from small. Now i see that they may still cancel each other but at least it's not as dumb as the scenario i described :) And other factors will stand (small, quality, momentum, low vol)

Is it possible to have both ? I guess you can create your own fund that does min variance with factor constraints (no neg. value) but i don't want to do that :)
I understand your conundrum. That is why Multifactor funds came out; because people holding the individual factor or tilt funds would have some issues like you've described.

However, it's not all bad. Say there's a you'd need to hold 20 shares of to be at market weight. The stock is low beta so EFAV holds 30 of them but it's growth, so ISCF holds only 10 of them. So you end up holding at market weight which makes sense since it has positive and negative qualities for you. That said, there are frictional costs because ISCF would sell while EFAV would buy. It's best to just hold still. That's the magic of the Multifactor fund.

Any ways, I'd recommend two things. Maybe ask hdas about it, he knows his Low Vol stuff and might have an answer.

The implication of the above numbers (and what I would personally do) is not worry about it. If EFAV continues to shift around, then your long term value exposure will be close to zero. There might be bad timing (growth-y when value outperforms and vice versa) but hopefully canceled out by good timing (growth-y when grown outperforms and vice versa). That's why I just look at the long term exposure and buy/hold based on those. I don't concern myself too much with whether my CURRENT exposure is correct.

BTW I have some number for long term factor exposure of all of these funds in case you're interested. It sounds like you're mostly worried about what they are right now so nvm in that case.

Cheers man

klaus14
Posts: 280
Joined: Sun Nov 25, 2018 7:43 pm

Re: Four International ETFs that are better diversifiers than VXUS

Post by klaus14 » Mon Oct 28, 2019 3:09 pm

305pelusa wrote:
Mon Oct 28, 2019 2:30 pm
klaus14 wrote:
Mon Oct 28, 2019 2:08 pm
305pelusa wrote:
Mon Oct 28, 2019 1:28 pm
klaus14 wrote:
Mon Oct 28, 2019 1:12 pm
305pelusa wrote:
Mon Oct 28, 2019 1:02 pm


AFAIK, this fund isn't applying any value or non-value orientations. Its methodology is purely about volatility. Based on their low volatility methodology, they choose stocks.

If you look at the stocks that methodology chose these past 10 years, and then apply a value test to it (whether that's by calculating it with their own definition you provided, or using FF regression), you find that they tended to be growth stocks.

If you apply the same value test to the stocks this methodology would've chosen from 2000-2010 however, they tended to be value-y stocks. Again, regardless of your definition used.

One might then make a reasonable conclusion that this isn't neither a growth or value fund. It's a min volatility fund whose exposure to other values will change over the time due to chance. You might reach other conclusions instead. I'm just giving you the bigger picture instead of just the last decade of the index.
i understand what you are saying. you apply factor regression based on let's say AQR data or Fema French data. However, in this instance, MSCI is playing the role of AQR. They don't need to estimate the factor loads by trying to fit history to some benchmark. They are the ones defining things, they can just compute the p/b and p/e of the underlying stocks. So if they are saying currently fund has negative value loading, we don't need to question that. I think HmL is just book value, i think MSCI definition makes more sense than just using book value.

But if you are saying that current negative load may change in future because this fund doesn't care about value, then of course i would agree. My question was about if current negative value makes it a bad companion to my other positive value loaded funds. i think the answer is yes. And to mitigate that my solution is to use multifactor funds (ISCF) for small cap space and min vol (EFAV) for large cap. in this way, i am getting both factor loadings.
Yes, we're on the same page.

Personally, I don't think the answer to that question is necessarily "yes". Are you looking to hold these funds long term or you'll just switch around based on the current loadings?

Ex: If in 2 years EFAV becomes large, it will "cancel" out ISCF. So then will you switch back to INTF ?

Obviously it depends on your goals. I assumed you're looking to buy and hold these long term. In which case I would place more weight on their long term factor exposure instead of whatever it happened to be on the day you purchased them.

Also, to be clear, if EFAV is growth-y right now, it will "cancel" out whatever value exposure you have, regardless of the fund (INTF or ISCF) used. The value premium does tend to be larger on smaller stocks but that's a bit of a second order effect. Is that what you're banking on?
Unfortunately ISCF value load is pretty weak: https://www.msci.com/documents/10199/8f ... 798db32f38

I guess what i am trying to do is:
I like min-variance idea and i want to add it to my portfolio. At the same time i am trying to avoid the situation like EFAV overweights stock X but INTF underweights stock X, so they cancel each other and i am paying extra fees for nothing.
I am inspired by VFMF construction: they divide universe into 3 equal parts (small, mid, large) and do factor optimization within each. In my case, i'll be getting min. vol. premium from large and value from small. Now i see that they may still cancel each other but at least it's not as dumb as the scenario i described :) And other factors will stand (small, quality, momentum, low vol)

Is it possible to have both ? I guess you can create your own fund that does min variance with factor constraints (no neg. value) but i don't want to do that :)
I understand your conundrum. That is why Multifactor funds came out; because people holding the individual factor or tilt funds would have some issues like you've described.

However, it's not all bad. Say there's a you'd need to hold 20 shares of to be at market weight. The stock is low beta so EFAV holds 30 of them but it's growth, so ISCF holds only 10 of them. So you end up holding at market weight which makes sense since it has positive and negative qualities for you. That said, there are frictional costs because ISCF would sell while EFAV would buy. It's best to just hold still. That's the magic of the Multifactor fund.

Any ways, I'd recommend two things. Maybe ask hdas about it, he knows his Low Vol stuff and might have an answer.

The implication of the above numbers (and what I would personally do) is not worry about it. If EFAV continues to shift around, then your long term value exposure will be close to zero. There might be bad timing (growth-y when value outperforms and vice versa) but hopefully canceled out by good timing (growth-y when grown outperforms and vice versa). That's why I just look at the long term exposure and buy/hold based on those. I don't concern myself too much with whether my CURRENT exposure is correct.

BTW I have some number for long term factor exposure of all of these funds in case you're interested. It sounds like you're mostly worried about what they are right now so nvm in that case.

Cheers man
Thanks. This all makes sense.

I guess if EFAV starts becoming more value-y this is a good thing for my setup, so no concern. if it starts covering more small cap stocks then that would increase the chances of conflicting with ISCF but you are saying even that is not a big deal.

But i guess you would agree that EFAV + ISCF is a better pairing than EFAV + INTF.

Post Reply