Vanguard ETFs, ranked by diversification benefit

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Topic Author
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

I was analyzing some ETFs today and ended up with a list of Vanguard ETFs and their diversification ratios.

The concept of diversification is not well understood by investors, I think, so I hope this will spur some thoughtful discussion.

Basically, diversification is the process of spreading the total amount risk in a portfolio more evenly across the sources of risk. A well-diversified portfolio allocates its risk across many independent sources as feasible.

Some investors confuse "diversification" (which is basically the balancing of risk) with "de-risking" (which is basically the reducing of risk). Improvements in diversification can increase, decrease, or leave unchanged the overall level of risk in a portfolio. Likewise diversification can increase, decrease, or leave unchanged the overall level of expected return in a portfolio. In this way, you can imagine a portfolio being supported by three legs or pillars: expected return, expected risk, and level of diversification. For any given level of expected return and risk, the investor should prefer a higher level of diversification to a lower level.

Mathematically, diversification effectively depends on the correlations of portfolio holdings and the variance (or volatility) of those holdings. Generally speaking, low correlations and higher variances make an asset a better diversifier. One way to easily estimate the level of portfolio diversification to compute the average volatility of the portfolio assets and compare that to the overall volatility of the combined portfolio. The higher the ratio, the more diversified the portfolio.

In short, the higher the diversification ratio the more diversified the portfolio. There are a lot of technical qualifications, but that's the gist of it.

What follows is a list of Vanguard ETFs (I only computed funds with inception dates in 2015 or earlier) with the diversification ratio computed in a 50/50 portfolio with Vanguard Total Stock Market ETF (VTI).

Code: Select all

Vanguard Extended Duration Trs ETF (EDV): 1.60
Vanguard Long-Term Treasury ETF (VGLT): 1.59
Vanguard Long-Term Bond ETF (BLV): 1.37
Vanguard Intmdt-Term Trs ETF (VGIT): 1.35
Vanguard Utilities ETF (VPU): 1.30
Vanguard Interm-Term Bond ETF (BIV): 1.29
Vanguard Tax-Exempt Bond ETF (VTEB): 1.26
Vanguard Total Bond Market ETF (BND): 1.24
Vanguard Long-Term Corporate Bd ETF (VCLT): 1.23
Vanguard Mortgage-Backed Secs ETF (VMBS): 1.21
Vanguard Interm-Term Corp Bd ETF (VCIT): 1.21
Vanguard Total International Bond ETF (BNDX): 1.20
Vanguard Emerging Mkts Govt Bd ETF (VWOB): 1.20
Vanguard Short-Term Bond ETF (BSV): 1.13
Vanguard Consumer Staples ETF (VDC): 1.13
Vanguard Real Estate ETF (VNQ): 1.11
Vanguard Short-Term Treasury ETF (VGSH): 1.11
Vanguard Global ex-US Real Est ETF (VNQI): 1.10
Vanguard Short-Term Corporate Bond ETF (VCSH): 1.10
Vanguard FTSE Emerging Markets ETF (VWO): 1.10
Vanguard Communication Services ETF (VOX): 1.09
Vanguard Short-Term Infl-Prot Secs ETF (VTIP): 1.08
Vanguard Energy ETF (VDE): 1.07
Vanguard FTSE Europe ETF (VGK): 1.07
Vanguard FTSE All-Wld ex-US SmCp ETF (VSS): 1.05
Vanguard FTSE All-Wld ex-US ETF (VEU): 1.05
Vanguard Total International Stock ETF (VXUS): 1.05
Vanguard Health Care ETF (VHT): 1.05
Vanguard FTSE Pacific ETF (VPL): 1.05
Vanguard Information Technology ETF (VGT): 1.04
Vanguard Financials ETF (VFH): 1.04
Vanguard FTSE Developed Markets ETF (VEA): 1.04
Vanguard Russell 2000 Value ETF (VTWV): 1.04
Vanguard S&P Small-Cap 600 Value ETF (VIOV): 1.04
Vanguard S&P Small-Cap 600 ETF (VIOO): 1.03
Vanguard S&P Small-Cap 600 Growth ETF (VIOG): 1.03
Vanguard Materials ETF (VAW): 1.03
Vanguard Russell 2000 ETF (VTWO): 1.03
Vanguard S&P Mid-Cap 400 Value ETF (IVOV): 1.02
Vanguard Russell 2000 Growth ETF (VTWG): 1.02
Vanguard Small-Cap Value ETF (VBR): 1.02
Vanguard Small-Cap Growth ETF (VBK): 1.02
Vanguard Industrials ETF (VIS): 1.02
Vanguard Consumer Discretionary ETF (VCR): 1.02
Vanguard Small-Cap ETF (VB): 1.02
Vanguard S&P 500 Growth ETF (VOOG): 1.01
Vanguard Mega Cap Value ETF (MGV): 1.01
Vanguard Mega Cap Growth ETF (MGK): 1.01
Vanguard Dividend Appreciation ETF (VIG): 1.01
Vanguard S&P Mid-Cap 400 Growth ETF (IVOG): 1.01
Vanguard High Dividend Yield ETF (VYM): 1.01
Vanguard S&P Mid-Cap 400 ETF (IVOO): 1.01
Vanguard Extended Market ETF (VXF): 1.01
Vanguard S&P 500 Value ETF (VOOV): 1.01
Vanguard Total World Stock ETF (VT): 1.01
Vanguard Mid-Cap Value ETF (VOE): 1.01
Vanguard Russell 1000 Growth ETF (VONG): 1.01
Vanguard Value ETF (VTV): 1.01
Vanguard Growth ETF (VUG): 1.01
Vanguard Mid-Cap Growth ETF (VOT): 1.01
Vanguard Russell 1000 Value ETF (VONV): 1.01
Vanguard Mid-Cap ETF (VO): 1.01
Vanguard Russell 3000 ETF (VTHR): 1.00
Vanguard Mega Cap ETF (MGC): 1.00
Vanguard Large-Cap ETF (VV): 1.00
Vanguard Russell 1000 ETF (VONE): 1.00
Vanguard S&P 500 ETF (VOO): 1.00
Because each line represents a pair of assets in a portfolio, as you built a portfolio you'd have to recompute each mix of assets. Portfolio Visualizer used to do this but not longer does. It's not hard to do in Excel or any spreadsheet program that has a solver function.

Some takeaways:

1) The power of long-term bonds as a diversifier for equities is WAY more powerful than most people realize. The top two funds (EDV and VGLT) are far ahead of the rest of the pack, highlighting the importance of including long-term bonds in the portfolio if your investment horizon is sufficiently long.

2) Utility stocks are an unsung hero of diversification. They have sufficiently low correlations with other stock sectors and sufficiently high variance to be powerful, especially when you consider that most broad indexes hold very little utility stocks. Even a 10% allocation to a utility sector index ETF can improve the diversification of a 3- or 4-fund portfolio. Another 10% in small-cap value helps too.

I saw another interesting phenomenon when I plotted each fund on an X-Y graph: correlation with VTI is on the X axis and diversification level is on the Y axis.

Image

The equity funds (mostly in the lower right in blue, with correlations over 0.50) and the long-term bond funds (mostly the blue dots with correlations below 0.40) ended up on an "efficient frontier" of sorts with a polynomial shape. I added a trend line to illustrate it, and the polynomial is related to the formula for covariance.

The short, intermediate, and total bond funds (illustrated in orange) generally fall well below this frontier. Despite have a low correlation with equities, they have insufficient variance to be effective diversifiers. This a way of visualizing the well-described fact that adding longer-term bonds, which many people think of as riskier, to an equity portfolio can actually produce an overall portfolio variance that is lower than if short-term bonds were used in the same proportion.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
MotoTrojan
Posts: 11259
Joined: Wed Feb 01, 2017 7:39 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by MotoTrojan »

If an investor’s horizon is long enough for 20-30 year bond funds, why would you suggest holding any bond at all?

I’m convinced that long treasuries will be the first Bonds I add to my core portfolio but I’m unsure about when to make that transition. Very risk-tolerant.
User avatar
Topic Author
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

MotoTrojan wrote: Mon May 20, 2019 4:32 pm If an investor’s horizon is long enough for 20-30 year bond funds, why would you suggest holding any bond at all?
I certainly agree that asset allocation, especially during accumulation, should primarily be driven by risk tolerance.

That said, the joint life expectancy of a married couple at age 63 is still something like 28 years or so. My guess is that most such couples would choose to be less than 100% in stocks, and their investment horizon is definitely long enough to warrant putting most or perhaps even all of whatever bond allocation they settle on into long-term bonds or bond funds.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
MotoTrojan
Posts: 11259
Joined: Wed Feb 01, 2017 7:39 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by MotoTrojan »

vineviz wrote: Mon May 20, 2019 5:10 pm
MotoTrojan wrote: Mon May 20, 2019 4:32 pm If an investor’s horizon is long enough for 20-30 year bond funds, why would you suggest holding any bond at all?
I certainly agree that asset allocation, especially during accumulation, should primarily be driven by risk tolerance.

That said, the joint life expectancy of a married couple at age 63 is still something like 28 years or so. My guess is that most such couples would choose to be less than 100% in stocks, and their investment horizon is definitely long enough to warrant putting most or perhaps even all of whatever bond allocation they settle on into long-term bonds or bond funds.
Thanks, thinking more along the lines of someone closer to 28 years of age, rather than 28 years of time horizon. The tinkerer in me gets excited at the prospect of adding some EDV or TLT, but then I ask what I am I trying to accomplish.

Perhaps if I had a 7-figure windfall I would want to de-risk some but still maintain a long-term portfolio. Suppose I am A-okay at the moment.
pascalwager
Posts: 2327
Joined: Mon Oct 31, 2011 8:36 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by pascalwager »

MotoTrojan wrote: Mon May 20, 2019 5:39 pm
vineviz wrote: Mon May 20, 2019 5:10 pm
MotoTrojan wrote: Mon May 20, 2019 4:32 pm If an investor’s horizon is long enough for 20-30 year bond funds, why would you suggest holding any bond at all?
I certainly agree that asset allocation, especially during accumulation, should primarily be driven by risk tolerance.

That said, the joint life expectancy of a married couple at age 63 is still something like 28 years or so. My guess is that most such couples would choose to be less than 100% in stocks, and their investment horizon is definitely long enough to warrant putting most or perhaps even all of whatever bond allocation they settle on into long-term bonds or bond funds.
Thanks, thinking more along the lines of someone closer to 28 years of age, rather than 28 years of time horizon. The tinkerer in me gets excited at the prospect of adding some EDV or TLT, but then I ask what I am I trying to accomplish.

Perhaps if I had a 7-figure windfall I would want to de-risk some but still maintain a long-term portfolio. Suppose I am A-okay at the moment.
If you were willing to be guided by the Estrada studies, you might decide to use a rising equity glidepath beginning at say age 25 with an 20/80 AA. At retirement (having gradually become 80/20 over the decades), the glidepath then becomes declining equity.

But Estrada doesn't specify the type of bonds.
VT 60% / VFSUX 20% / TIPS 20%
Typ997S
Posts: 98
Joined: Fri Apr 06, 2012 12:36 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by Typ997S »

To the OP: Thanks for this post...very interesting. Could you elaborate on why you state adding 10% small value is helpful when your numbers show 1.02 for all the Vanguard Small Cap funds? I understand Utilities when the factor is 1.3, but not small cap at 1.02. I'm a very interested observer since I own a lot of small cap (VB and VBR) and no utilities!
MotoTrojan
Posts: 11259
Joined: Wed Feb 01, 2017 7:39 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by MotoTrojan »

pascalwager wrote: Mon May 20, 2019 5:51 pm
MotoTrojan wrote: Mon May 20, 2019 5:39 pm
vineviz wrote: Mon May 20, 2019 5:10 pm
MotoTrojan wrote: Mon May 20, 2019 4:32 pm If an investor’s horizon is long enough for 20-30 year bond funds, why would you suggest holding any bond at all?
I certainly agree that asset allocation, especially during accumulation, should primarily be driven by risk tolerance.

That said, the joint life expectancy of a married couple at age 63 is still something like 28 years or so. My guess is that most such couples would choose to be less than 100% in stocks, and their investment horizon is definitely long enough to warrant putting most or perhaps even all of whatever bond allocation they settle on into long-term bonds or bond funds.
Thanks, thinking more along the lines of someone closer to 28 years of age, rather than 28 years of time horizon. The tinkerer in me gets excited at the prospect of adding some EDV or TLT, but then I ask what I am I trying to accomplish.

Perhaps if I had a 7-figure windfall I would want to de-risk some but still maintain a long-term portfolio. Suppose I am A-okay at the moment.
If you were willing to be guided by the Estrada studies, you might decide to use a rising equity glidepath beginning at say age 25 with an 20/80 AA. At retirement (having gradually become 80/20 over the decades), the glidepath then becomes declining equity.

But Estrada doesn't specify the type of bonds.
Interesting. Not familiar but I can’t imagine how that comes out ahead.
Thesaints
Posts: 5108
Joined: Tue Jun 20, 2017 12:25 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by Thesaints »

vineviz wrote: Mon May 20, 2019 4:22 pm The short, intermediate, and total bond funds (illustrated in orange) generally fall well below this frontier. Despite have a low correlation with equities, they have insufficient variance to be effective diversifiers. This a way of visualizing the well-described fact that adding longer-term bonds, which many people think of as riskier, to an equity portfolio can actually produce an overall portfolio variance that is lower than if short-term bonds were used in the same proportion.
Can ≠ Will

Clearly, with 50% of my portfolio in long-term bonds (max volatility amongst bond and zero correlation to stocks) I might be able to counter a loss on my 50% in stocks better than if I used cash (zero volatility and zero correlation with stocks).
But zero correlation does not mean that bonds will move in an opposite directions as stocks. It is equally probable that my long-term bonds will move in the same direction as stocks, thus adding to my loss.

So, yes, volatile bonds can help me do better, or they can help me do worse. Overall, the portfolio volatility will be higher than if I used cash instead.
All of the above assuming that stocks-bonds correlation is zero, but with monetary policies presently being the principal driver for both, chances are correlation will be as positive as it has ever been, even more so on the downside.

With the currently flattish yield curve, it is very possible that investors will be better served with an equivalent risk cash+high yield mix than with long-term bonds. Possibly even cash only can end up being a superior choice.
columbia
Posts: 3023
Joined: Tue Aug 27, 2013 5:30 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by columbia »

pascalwager wrote: Mon May 20, 2019 5:51 pm
MotoTrojan wrote: Mon May 20, 2019 5:39 pm
vineviz wrote: Mon May 20, 2019 5:10 pm
MotoTrojan wrote: Mon May 20, 2019 4:32 pm If an investor’s horizon is long enough for 20-30 year bond funds, why would you suggest holding any bond at all?
I certainly agree that asset allocation, especially during accumulation, should primarily be driven by risk tolerance.

That said, the joint life expectancy of a married couple at age 63 is still something like 28 years or so. My guess is that most such couples would choose to be less than 100% in stocks, and their investment horizon is definitely long enough to warrant putting most or perhaps even all of whatever bond allocation they settle on into long-term bonds or bond funds.
Thanks, thinking more along the lines of someone closer to 28 years of age, rather than 28 years of time horizon. The tinkerer in me gets excited at the prospect of adding some EDV or TLT, but then I ask what I am I trying to accomplish.

Perhaps if I had a 7-figure windfall I would want to de-risk some but still maintain a long-term portfolio. Suppose I am A-okay at the moment.
If you were willing to be guided by the Estrada studies, you might decide to use a rising equity glidepath beginning at say age 25 with an 20/80 AA. At retirement (having gradually become 80/20 over the decades), the glidepath then becomes declining equity.

But Estrada doesn't specify the type of bonds.

Can you point me to a digestible summary of that?
Angst
Posts: 2968
Joined: Sat Jun 09, 2007 11:31 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by Angst »

This really is interesting. I'm not too surprised with the rankings but the utilities and a few others jumped out. I do have some difficulty though understanding your interpretation of the significance of the graphed results vs. VTI

By definition, you've already incorporated both the correlations and the variance of the different pairings into the diversification ratio which you've characterized as a hierarchy for identifying "a better diversifier", but in the graphed results you're deprecating the orange dots off to the left for lacking "variance" vs. parallel blue dots off to the right which have much higher correlations to VTI while holding essentially identical diversification ratios with the orange dots.

It seems to me that you're showing in the graphed results that in a 50:50 portfolio where one half is VTI, I would, for example, be better off pairing with VTI one of the following "equity" funds (Blue dots), all of which (except one which is close) have >.6 correlations with VTI:

Code: Select all

Vanguard Consumer Staples ETF (VDC):           1.13
Vanguard Real Estate ETF (VNQ):                1.11
Vanguard Global ex-US Real Est ETF (VNQI):     1.10
Vanguard FTSE Emerging Markets ETF (VWO):      1.10
Vanguard Communication Services ETF (VOX):     1.09
than pairing with VTI any one of these following ST Bond funds (Orange dots) which have correlations with VTI of roughly (my eyeballing) of (.35), (.21), .15 and .18

Code: Select all

Vanguard Short-Term Bond ETF (BSV):            1.13
Vanguard Short-Term Treasury ETF (VGSH):       1.11
Vanguard Short-Term Corporate Bond ETF (VCSH): 1.10
Vanguard Short-Term Infl-Prot Secs ETF (VTIP): 1.08
So when the market crashes, would I really prefer to have VTI paired 50:50 with one of these equity ETF's having correlations like .6 vs. any of the above bond funds having correlations ranging from (.35) to .18? I find that counter-intuitive, to say the least. Is that really what you mean?
pascalwager
Posts: 2327
Joined: Mon Oct 31, 2011 8:36 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by pascalwager »

columbia wrote: Mon May 20, 2019 6:51 pm
pascalwager wrote: Mon May 20, 2019 5:51 pm
MotoTrojan wrote: Mon May 20, 2019 5:39 pm
vineviz wrote: Mon May 20, 2019 5:10 pm
MotoTrojan wrote: Mon May 20, 2019 4:32 pm If an investor’s horizon is long enough for 20-30 year bond funds, why would you suggest holding any bond at all?
I certainly agree that asset allocation, especially during accumulation, should primarily be driven by risk tolerance.

That said, the joint life expectancy of a married couple at age 63 is still something like 28 years or so. My guess is that most such couples would choose to be less than 100% in stocks, and their investment horizon is definitely long enough to warrant putting most or perhaps even all of whatever bond allocation they settle on into long-term bonds or bond funds.
Thanks, thinking more along the lines of someone closer to 28 years of age, rather than 28 years of time horizon. The tinkerer in me gets excited at the prospect of adding some EDV or TLT, but then I ask what I am I trying to accomplish.

Perhaps if I had a 7-figure windfall I would want to de-risk some but still maintain a long-term portfolio. Suppose I am A-okay at the moment.
If you were willing to be guided by the Estrada studies, you might decide to use a rising equity glidepath beginning at say age 25 with an 20/80 AA. At retirement (having gradually become 80/20 over the decades), the glidepath then becomes declining equity.

But Estrada doesn't specify the type of bonds.

Can you point me to a digestible summary of that?
https://papers.ssrn.com/sol3/papers.cfm ... id=2557256

I don't want to sidetrack vineviz' thread, but Estrada says it works because you end up with the highest % of stocks when the portfolio is largest--near retirement. Arnott also came to the same conclusion with his own study.
VT 60% / VFSUX 20% / TIPS 20%
MotoTrojan
Posts: 11259
Joined: Wed Feb 01, 2017 7:39 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by MotoTrojan »

pascalwager wrote: Mon May 20, 2019 7:50 pm
columbia wrote: Mon May 20, 2019 6:51 pm
pascalwager wrote: Mon May 20, 2019 5:51 pm
MotoTrojan wrote: Mon May 20, 2019 5:39 pm
vineviz wrote: Mon May 20, 2019 5:10 pm

I certainly agree that asset allocation, especially during accumulation, should primarily be driven by risk tolerance.

That said, the joint life expectancy of a married couple at age 63 is still something like 28 years or so. My guess is that most such couples would choose to be less than 100% in stocks, and their investment horizon is definitely long enough to warrant putting most or perhaps even all of whatever bond allocation they settle on into long-term bonds or bond funds.
Thanks, thinking more along the lines of someone closer to 28 years of age, rather than 28 years of time horizon. The tinkerer in me gets excited at the prospect of adding some EDV or TLT, but then I ask what I am I trying to accomplish.

Perhaps if I had a 7-figure windfall I would want to de-risk some but still maintain a long-term portfolio. Suppose I am A-okay at the moment.
If you were willing to be guided by the Estrada studies, you might decide to use a rising equity glidepath beginning at say age 25 with an 20/80 AA. At retirement (having gradually become 80/20 over the decades), the glidepath then becomes declining equity.

But Estrada doesn't specify the type of bonds.

Can you point me to a digestible summary of that?
https://papers.ssrn.com/sol3/papers.cfm ... id=2557256

I don't want to sidetrack vineviz' thread, but Estrada says it works because you end up with the highest % of stocks when the portfolio is largest--near retirement. Arnott also came to the same conclusion with his own study.
The same is true of 80/20 until retirement.
User avatar
Topic Author
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

MotoTrojan wrote: Mon May 20, 2019 5:39 pm [Thanks, thinking more along the lines of someone closer to 28 years of age, rather than 28 years of time horizon. The tinkerer in me gets excited at the prospect of adding some EDV or TLT, but then I ask what I am I trying to accomplish.
I had no bonds when I was 28 either, and I slept like a baby.

If I could go back in time, knowing what I know now, I think a good strategy would be to start buying individual 30-year zero-coupon Treasuries at age 35: maybe 5% of gross income each year. By the time you turn 65 you have a 30 year ladder already built, providing a nominal baseline for retirement income.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
MotoTrojan
Posts: 11259
Joined: Wed Feb 01, 2017 7:39 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by MotoTrojan »

vineviz wrote: Mon May 20, 2019 8:09 pm
MotoTrojan wrote: Mon May 20, 2019 5:39 pm [Thanks, thinking more along the lines of someone closer to 28 years of age, rather than 28 years of time horizon. The tinkerer in me gets excited at the prospect of adding some EDV or TLT, but then I ask what I am I trying to accomplish.
I had no bonds when I was 28 either, and I slept like a baby.

If I could go back in time, knowing what I know now, I think a good strategy would be to start buying individual 30-year zero-coupon Treasuries at age 35: maybe 5% of gross income each year. By the time you turn 65 you have a 30 year ladder already built, providing a nominal baseline for retirement income.
Thanks. Main advantages over a fund like EDV being the lack of fees and easier ability to roll off your maturity as you approach retirement?
User avatar
Topic Author
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

Thesaints wrote: Mon May 20, 2019 6:49 pm But zero correlation does not mean that bonds will move in an opposite directions as stocks. It is equally probable that my long-term bonds will move in the same direction as stocks, thus adding to my loss.
The correlation of long-term Treasuries with stocks is lower than zero, by a significant amount. But it IS true there are no guarantees in investing.


Thesaints wrote: Mon May 20, 2019 6:49 pmOverall, the portfolio volatility will be higher than if I used cash instead.
Not necessarily, and not usually. Thoughtful diversification is more powerful than most people assume it is.


Over the past 22 years, for instance, a 60/40 portfolio with ultrashort Treasuries had a higher standard deviation than a 60/40 portfolio with long-term Treasuries..

https://www.portfoliovisualizer.com/bac ... 0&total3=0
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
johndough
Posts: 19
Joined: Tue Dec 30, 2008 9:24 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by johndough »

Very interesting and thanks for sharing. Would be interesting to see what set of portfolio weights results in the max diversification ratio. I'd probably add LTPZ to the funds as well.
AHTFY
Posts: 173
Joined: Mon Jul 23, 2018 2:33 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by AHTFY »

How about REITs?
pdavi21
Posts: 1296
Joined: Sat Jan 30, 2016 3:04 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by pdavi21 »

If I get 55 people in a room and average their height, I get 5.6'.

If I then find another 45 people from another country and average their heights, I get 5.5'.

I then pick 3 out of the original 55 people. Their average height is 5.2'.

Then I pick one person, he is over 6.5'.

Then I pick 2 out of the 45. They are both under 5'.

Do you see what I'm getting at here?
"We spend a great deal of time studying history, which, let's face it, is mostly the history of stupidity." -Stephen Hawking
fennewaldaj
Posts: 1097
Joined: Sun Oct 22, 2017 11:30 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by fennewaldaj »

Emerging market bonds have a decent diversification ratio. I would assume that emerging local currency would be higher since they are a fair bit more volatile?
272 Sheep
Posts: 158
Joined: Thu Mar 26, 2009 4:14 pm
Location: Hooksett, NH

Re: Vanguard ETFs, ranked by diversification benefit

Post by 272 Sheep »

OP,
In likely rising interest rate environment, not sure that having (especially significant amount) of long bonds is a good idea, especially a person
like me who is early in his retirement. Sharp rate climb will give significant and proportional loss to long-bond value. Aren't long-bonds closely
correlated with equities so they act very similarly in a portfolio? I want my bonds to act differently (if possible) than my stocks.
Swedroe always says, take risk in stocks, safety in bonds.

If you are a younger investor (have time on your side) and confident and have the discipline to follow through with your strategy, that "could" work very well for you. Especially, that you are likely holding far lower percentage of bonds. I (have less time on my side) am fairly sure long-bond mix is not for me. Unlikely I will trade in short-term and intermediates for long-bonds in my situation.
Carl W.
Thesaints
Posts: 5108
Joined: Tue Jun 20, 2017 12:25 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by Thesaints »

vineviz wrote: Mon May 20, 2019 8:23 pm
Thesaints wrote: Mon May 20, 2019 6:49 pm But zero correlation does not mean that bonds will move in an opposite directions as stocks. It is equally probable that my long-term bonds will move in the same direction as stocks, thus adding to my loss.
The correlation of long-term Treasuries with stocks is lower than zero, by a significant amount. But it IS true there are no guarantees in investing.


Thesaints wrote: Mon May 20, 2019 6:49 pmOverall, the portfolio volatility will be higher than if I used cash instead.
Not necessarily, and not usually. Thoughtful diversification is more powerful than most people assume it is.


Over the past 22 years, for instance, a 60/40 portfolio with ultrashort Treasuries had a higher standard deviation than a 60/40 portfolio with long-term Treasuries..

https://www.portfoliovisualizer.com/bac ... 0&total3=0
But only because in the past 22 years it turned out that the zero correlation generated inverse trends. That is not necessarily true in the future.
That is also what makes you conclude that long-term bonds are negatively correlated to stocks. It is simply what has happened in the recent past (in fact a past that excludes the most recent years), but there is no reason to expect it to take place again in the near future. In fact, we should probably expect the opposite: higher correlation than usual and in a larger measure for longer maturities.
AlohaJoe
Posts: 6609
Joined: Mon Nov 26, 2007 1:00 pm
Location: Saigon, Vietnam

Re: Vanguard ETFs, ranked by diversification benefit

Post by AlohaJoe »

pdavi21 wrote: Mon May 20, 2019 9:55 pm Do you see what I'm getting at here?
No. Could you just explain directly instead of offering an analogy?
AlohaJoe
Posts: 6609
Joined: Mon Nov 26, 2007 1:00 pm
Location: Saigon, Vietnam

Re: Vanguard ETFs, ranked by diversification benefit

Post by AlohaJoe »

pascalwager wrote: Mon May 20, 2019 5:51 pm But Estrada doesn't specify the type of bonds.
Bonds here means "long-term government bonds". The exact maturity varies depending on data available for each country but is generally either 10 or 20 years. It also isn't constant. It isn't like "for Belgium we use a maturity of 10 years from 1900-2015". So the maturity used for Belgian bonds from 1900-1956 (an average of whatever bonds of at least 5 years they could find data for) will be different than the maturity used for 1956-1987 (10 year bonds) and both of those will be different from the maturity used for 1987-onwards (20 year bonds).
User avatar
Topic Author
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

MotoTrojan wrote: Mon May 20, 2019 8:16 pm Thanks. Main advantages over a fund like EDV being the lack of fees and easier ability to roll off your maturity as you approach retirement?
Yes, a strategy like this naturally reduces duration to keep pace with the investment horizon.

Not everyone would want to manage a portfolio of individual nominal Treasury bonds (and maybe in the later years TIPS), as it is a little more work to set up and track. But once set up, there'd be less follow-up management.
"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: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

fennewaldaj wrote: Mon May 20, 2019 11:45 pm Emerging market bonds have a decent diversification ratio. I would assume that emerging local currency would be higher since they are a fair bit more volatile?
Yes. Vanguard doesn't currently have a an EM local currency bond fund, but a similar ETF (e.g. EMLC, EBND, or LEMB) would be somewhat better diversifier than a currency-hedged fund or a global bond fund.

I use LEMB (iShares J.P. Morgan EM Local Currency Bond ETF) in my portfolio, but because of the volatility and high yield (7.68% now, I think) I treat it as part of my equity allocation rather than my fixed income allocation.

And because the EM local currency bond index doesn't currently include China, a mix of EM equity and EM local currency bond has more geographic diversity than EM equity alone.
"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: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

Thesaints wrote: Tue May 21, 2019 2:09 am
vineviz wrote: Mon May 20, 2019 8:23 pm Over the past 22 years, for instance, a 60/40 portfolio with ultrashort Treasuries had a higher standard deviation than a 60/40 portfolio with long-term Treasuries..
But only because in the past 22 years it turned out that the zero correlation generated inverse trends. That is not necessarily true in the future.
I'd argue it's more accurate to say that the correlation between stocks and treasury bonds has been negative for the past 30 years. You are right to observe that the correlation has been positive in the past, and could possibly be positive in the future. I think there are economic and policy factors that make that unlikely, but clearly I could be wrong.

On the other hand, strictly negative correlations aren't actually necessary for the diversification power of longer duration bonds to kick in.

I think there are significant flaws in our pre-1980 return series for long-term bonds, for reasons I don't think we need to dive into here, but leaving that aside I took a look back to 1950 to include a long period (1965 to 1998) when the correlation between long bonds and stocks was positive.

From 1950 to 2018 there were 15 periods where stocks had drawdowns in excess of 10%, and the average such period had a drawdown of 23%.

A 60/40 portfolio using cash as the fixed income holding dropped an average of 12.7% during those same drawdowns, whereas a 60/40 portfolio using long term treasuries dropped an average of 13.0%. I'd argue that the difference in volatility is pretty subtle and one that most investors wouldn't notice if you they are paying attention to the portfolio as a whole (as they should)

The fact that the second portfolio had considerably higher returns with roughly the same volatility is pretty much a textbook example of how diversification works.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Thesaints
Posts: 5108
Joined: Tue Jun 20, 2017 12:25 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by Thesaints »

Yes, but now (not in the 50’s, 60’s, or the 90’s), yield curve is as flat as it can be and monetary policies have set how much money is on the table to be invested in financial markets, beyond what investors can decide.
It means expected return from cash and long-term treasuries won’t be as different as in the past and correlation between stocks and bonds will be higher than in the past.

There are an evident reasons why, as you observe, in the past long-term treasuries have yielded better portfolio risk-normalized returns than cash, but those reasons are not there today, or are there but much weakened.

I’m throwing here my forecast: a suitable mix of cash and HY bonds will outperform on a risk-normalized basis a portfolio that instead uses LT treasuries as its bonds anchor. Given the size of the HY market this is not something institutional investors may pursue extensively, but for individual investors the story is different. Another of those not frequent opportunities where the small fish has an advantage over the big fish.
pdavi21
Posts: 1296
Joined: Sat Jan 30, 2016 3:04 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by pdavi21 »

AlohaJoe wrote: Tue May 21, 2019 2:51 am
pdavi21 wrote: Mon May 20, 2019 9:55 pm Do you see what I'm getting at here?
No. Could you just explain directly instead of offering an analogy?
The correlation is related to a few factors:
1. How diversified is the stock fund? Fewer holdings introduce a level of randomness. (i.e. the one guy who is 6'6")
If we replace some of the smaller funds like VIOG with 100% AAPL, we find that AAPL is likely to have a lower correlation despite being a larger portion of VTI.
2. How much of VTI is represented by the stock fund. (~0) for INTL, (<7%) for small cap? If you test an SP 500 Fund, it's going to offer a pretty low benefit because it represents most of VTI. This is the 2 or 3 people measured and would apply to funds like VSS and VB. VOO is like measuring 40+ people.
3. Random results from back-testing. If you measure one person that happens to be 5.6', it doesn't mean that that choice was poor for diversification; it just means you got unlucky. (EDIT: in this case my analogy fails unless people grow and shrink drastically over time).

An extra point I'd like to make is that the analysis is mostly useless for stock funds because any of the hot plays over the last decade would have artificially shown a high "diversification benefit" due to the above factors. Even going into factors could yield somewhat misleading results (although they seem to be glaringly absent) because the factor funds have fewer holdings and a lower piece of VTI than their index.
"We spend a great deal of time studying history, which, let's face it, is mostly the history of stupidity." -Stephen Hawking
Thesaints
Posts: 5108
Joined: Tue Jun 20, 2017 12:25 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by Thesaints »

pdavi21 wrote: Tue May 21, 2019 11:17 am 3. Random results from back-testing. If you measure one person that happens to be 5.6', it doesn't mean that that choice was poor for diversification; it just means you got unlucky.
The importance of the above statement can never be overstated. Past events are not necessarily all the events that could have happened. In fact, they might not even be the events most likely to have happened !
Backtesting with no theoretical backup is one of the most dangerous things in life, not only in finance.
pdavi21
Posts: 1296
Joined: Sat Jan 30, 2016 3:04 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by pdavi21 »

272 Sheep wrote: Mon May 20, 2019 11:51 pm OP,
In likely rising interest rate environment, not sure that having (especially significant amount) of long bonds is a good idea, especially a person
like me who is early in his retirement. Sharp rate climb will give significant and proportional loss to long-bond value. Aren't long-bonds closely
correlated with equities so they act very similarly in a portfolio? I want my bonds to act differently (if possible) than my stocks.
Swedroe always says, take risk in stocks, safety in bonds.

If you are a younger investor (have time on your side) and confident and have the discipline to follow through with your strategy, that "could" work very well for you. Especially, that you are likely holding far lower percentage of bonds. I (have less time on my side) am fairly sure long-bond mix is not for me. Unlikely I will trade in short-term and intermediates for long-bonds in my situation.
Carl W.
I disagree with a few of the comments in here, but implementation is spot on, in my opinion.
A retiree should have very little long bonds. Funds like BND/BNDX already carry some long bonds.
A young person with a long investment horizon and a very high risk tolerance would probably be fine in investing 0-20% of their portfolio in 100% long term bonds and 80-100% in stocks. As rates rise (if they do), they'll keep buying.
"We spend a great deal of time studying history, which, let's face it, is mostly the history of stupidity." -Stephen Hawking
MichCPA
Posts: 884
Joined: Fri Jul 06, 2018 9:06 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by MichCPA »

Thesaints wrote: Tue May 21, 2019 11:22 am
pdavi21 wrote: Tue May 21, 2019 11:17 am 3. Random results from back-testing. If you measure one person that happens to be 5.6', it doesn't mean that that choice was poor for diversification; it just means you got unlucky.
The importance of the above statement can never be overstated. Past events are not necessarily all the events that could have happened. In fact, they might not even be the events most likely to have happened !
Backtesting with no theoretical backup is one of the most dangerous things in life, not only in finance.
Plus using the past to draw conclusions about LT bonds is highly suspect because we are in an environment where there is virtually 0 chance that interest rates will drop as far as they have over the past 35 years. I don't see people signing up for a -6% LT bond anytime soon.
chisey
Posts: 277
Joined: Mon Apr 16, 2007 10:47 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by chisey »

MichCPA wrote: Tue May 21, 2019 12:15 pm
Plus using the past to draw conclusions about LT bonds is highly suspect because we are in an environment where there is virtually 0 chance that interest rates will drop as far as they have over the past 35 years. I don't see people signing up for a -6% LT bond anytime soon.
I think that's an important observation for returns of LT bonds in isolation, but I'm not sure it has much to say about LT bonds as a diversifier to equities.

The rate on a 30-year treasury was 2.87% as of December 1, 2008. The rate on the 30-year treasury was also 2.87% on May 1, 2019. Interest rates have not gone down over the period.

In those 10.5 years, VTI and TLT (iShares LTT) had a correlation of -0.31, and a 50/50 combination of them had a diversification ratio of 1.70.
Angst
Posts: 2968
Joined: Sat Jun 09, 2007 11:31 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by Angst »

Bump...

Vineviz, I'm disappointed you haven't commented yet on my post (below) from last night. Perhaps it just got past you in the course of things.
Angst wrote: Mon May 20, 2019 7:07 pm This really is interesting. I'm not too surprised with the rankings but the utilities and a few others jumped out. I do have some difficulty though understanding your interpretation of the significance of the graphed results vs. VTI

By definition, you've already incorporated both the correlations and the variance of the different pairings into the diversification ratio which you've characterized as a hierarchy for identifying "a better diversifier", but in the graphed results you're deprecating the orange dots off to the left for lacking "variance" vs. parallel blue dots off to the right which have much higher correlations to VTI while holding essentially identical diversification ratios with the orange dots.

It seems to me that you're showing in the graphed results that in a 50:50 portfolio where one half is VTI, I would, for example, be better off pairing with VTI one of the following "equity" funds (Blue dots), all of which (except one which is close) have >.6 correlations with VTI:

Code: Select all

Vanguard Consumer Staples ETF (VDC):           1.13
Vanguard Real Estate ETF (VNQ):                1.11
Vanguard Global ex-US Real Est ETF (VNQI):     1.10
Vanguard FTSE Emerging Markets ETF (VWO):      1.10
Vanguard Communication Services ETF (VOX):     1.09
than pairing with VTI any one of these following ST Bond funds (Orange dots) which have correlations with VTI of roughly (my eyeballing) of (.35), (.21), .15 and .18

Code: Select all

Vanguard Short-Term Bond ETF (BSV):            1.13
Vanguard Short-Term Treasury ETF (VGSH):       1.11
Vanguard Short-Term Corporate Bond ETF (VCSH): 1.10
Vanguard Short-Term Infl-Prot Secs ETF (VTIP): 1.08
So when the market crashes, would I really prefer to have VTI paired 50:50 with one of these equity ETF's having correlations like .6 vs. any of the above bond funds having correlations ranging from (.35) to .18? I find that counter-intuitive, to say the least. Is that really what you mean?
While some people have since been arguing for ST or IT bonds in certain circumstances over your preference for LT, what attracts my attention with respect to this part of the discussion even more is that based on your graphic in the OP you'd even prefer Vanguard Consumer Staples ETF (VDC) over Vanguard Short-Term Bond ETF (BSV). I'm surprised this isn't being discussed as well.
User avatar
Topic Author
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

Thesaints wrote: Tue May 21, 2019 11:04 am Yes, but now (not in the 50’s, 60’s, or the 90’s), yield curve is as flat as it can be and monetary policies have set how much money is on the table to be invested in financial markets, beyond what investors can decide.
It means expected return from cash and long-term treasuries won’t be as different as in the past and correlation between stocks and bonds will be higher than in the past.
Of course expected returns will matter to investors, but they are separate from the question of diversification since returns don't factor into it one way or the other.

As for the hypothesis that current yield curve and/or yield structure is predictive of future correlations, the evidence I've seen don't support that.
"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: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

MichCPA wrote: Tue May 21, 2019 12:15 pm Plus using the past to draw conclusions about LT bonds is highly suspect because we are in an environment where there is virtually 0 chance that interest rates will drop as far as they have over the past 35 years. I don't see people signing up for a -6% LT bond anytime soon.
The diversification benefits of long-term bonds don't depend on their expected returns, and I personally think investors avoid thinking they can successfully time bond market.

Bond yields are what they are because the market participants have in aggregate taken into account their expectations about what kind of "environment" we are in.

Some Bogleheads think that bond markets are inefficient and that they are smarter than the collective wisdom of the world's investors. To them I say "good luck".

Everyone else should pick a bond duration based on their investment horizon and call it a day.
"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: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

Angst wrote: Tue May 21, 2019 1:01 pm Vineviz, I'm disappointed you haven't commented yet on my post (below) from last night. Perhaps it just got past you in the course of things.
Sorry, I did indeed overlook this post.
Angst wrote: Mon May 20, 2019 7:07 pm So when the market crashes, would I really prefer to have VTI paired 50:50 with one of these equity ETF's having correlations like .6 vs. any of the above bond funds having correlations ranging from (.35) to .18? I find that counter-intuitive, to say the least. Is that really what you mean?
Solely from a diversification standpoint, you should be indifferent between two choices that have the same diversification ratio. Now, the data I presented was for simple 50/50 pairs of ETFs and virtually one of us have only two funds in our portfolios so we are operating at some level of simplification.

The diversification ratio is effectively a joint outpour of correlation and variance. Since we're taking VTI as a fixed partner, we can inver that if two other ETFs have the same diversification ratio but different correlations then they must also have different variances. Higher variances produce higher diversification ratios for a give correlation, so we can also infer that if one ETF has the same diversification ratio as another ETF but a lower correlation then the first ETF must have a HIGHER variance. Because markets are roughly mean-variance efficient, we can reasonable assume the the asset with higher variance has a higher expected return.

In other words, we can (I think) reasonably infer that a VTI/ETF pair on the blue line will have higher risk-adjusted return than a VTI/ETF pair having the same diversification ratio that is below the blue line.

That's a mouthful.

Angst wrote: Mon May 20, 2019 7:07 pmWhile some people have since been arguing for ST or IT bonds in certain circumstances over your preference for LT, what attracts my attention with respect to this part of the discussion even more is that based on your graphic in the OP you'd even prefer Vanguard Consumer Staples ETF (VDC) over Vanguard Short-Term Bond ETF (BSV). I'm surprised this isn't being discussed as well.
Me too, though in fairness diversification is just one dimension of portfolio optimization. Many investors (especially the relentlessly conservative ones that a forum like this tends to attract) are much more interested in de-risking their portfolio than in diversifying it. Adding a short-term bond fund will reduce the portfolio volatility quickly, even if it actually isn't diversifying the portfolio very much.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
acegolfer
Posts: 3029
Joined: Tue Aug 25, 2009 9:40 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by acegolfer »

vineviz wrote: Mon May 20, 2019 4:22 pm For any given level of expected return and risk, the investor should prefer a higher level of diversification to a lower level.
Can you explain the above? If 2 portfolios have the same E(r) and the same risks, why does the level of diversification matter?
dcb
Posts: 148
Joined: Wed Sep 26, 2007 12:44 pm
Location: Southwest Michigan

Re: Vanguard ETFs, ranked by diversification benefit

Post by dcb »

pdavi21 wrote: Mon May 20, 2019 9:55 pm If I get 55 people in a room and average their height, I get 5.6'.

If I then find another 45 people from another country and average their heights, I get 5.5'.

I then pick 3 out of the original 55 people. Their average height is 5.2'.

Then I pick one person, he is over 6.5'.

Then I pick 2 out of the 45. They are both under 5'.

Do you see what I'm getting at here?
delete
MichCPA
Posts: 884
Joined: Fri Jul 06, 2018 9:06 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by MichCPA »

vineviz wrote: Tue May 21, 2019 2:28 pm
MichCPA wrote: Tue May 21, 2019 12:15 pm Plus using the past to draw conclusions about LT bonds is highly suspect because we are in an environment where there is virtually 0 chance that interest rates will drop as far as they have over the past 35 years. I don't see people signing up for a -6% LT bond anytime soon.
The diversification benefits of long-term bonds don't depend on their expected returns, and I personally think investors avoid thinking they can successfully time bond market.

Bond yields are what they are because the market participants have in aggregate taken into account their expectations about what kind of "environment" we are in.

Some Bogleheads think that bond markets are inefficient and that they are smarter than the collective wisdom of the world's investors. To them I say "good luck".

Everyone else should pick a bond duration based on their investment horizon and call it a day.
This isn't a debate about market efficiency though. Its a debate about whether a low correlation to stocks makes LT bonds worth owning. I would furthermore point out that correlation doesn't by itself say much about risk. For instance Bitcoin has a .12 correlation to the stock market, but I don't think anyone says its safe. I would also like to point out that the S&P 500 total return on 10k over the past 10 years is about 40k and EDV has a return of 15K. Its almost inherent on that alone that those two things aren't highly correlated. It also means that LT bonds had a terrible return, and given the current rate of interest, there is a pretty narrow road to a good return. Interest rates seem to me to be a far preferable lens with which to view bonds, because they are a causative factor, not a correlation which can be muddied by all sorts of variables known and unknown.
MichCPA
Posts: 884
Joined: Fri Jul 06, 2018 9:06 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by MichCPA »

acegolfer wrote: Tue May 21, 2019 3:32 pm
vineviz wrote: Mon May 20, 2019 4:22 pm For any given level of expected return and risk, the investor should prefer a higher level of diversification to a lower level.
Can you explain the above? If 2 portfolios have the same E(r) and the same risks, why does the level of diversification matter?
+1

Especially since diversification is meant to eliminate idiosyncratic risk. I am assuming what this was meant to say is that for a given return, you should minimize risk. (IE don't take extra risk if you aren't paid more).
User avatar
Topic Author
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

acegolfer wrote: Tue May 21, 2019 3:32 pm
vineviz wrote: Mon May 20, 2019 4:22 pm For any given level of expected return and risk, the investor should prefer a higher level of diversification to a lower level.
Can you explain the above? If 2 portfolios have the same E(r) and the same risks, why does the level of diversification matter?
That question could take a book to answer fully, but I think the most salient points are these:

1) The expected return of a portfolio is calculated as the weighted average expected return of each holding, and those returns must be computed using an arithmetic mean. But arithmetic mean is only one of the three Pythagorean averages: geometric mean and harmonic mean are the other two. The compounded growth of a portfolio is more closely described by the geometric mean, and the sustainable withdrawal rate of a portfolio is more closely described by the harmonic mean. Confusingly, as a result, two portfolios with similar expected returns and expected variance will NOT necessarily produce the same expected outcome.

2) On a related note, we generally assume that investment returns are normally distributed and they mostly are. In a normal distribution the first two moments (mean and variance) are sufficient to describe the distribution, with investors generally favoring - all else equal - more of the former and less of the latter. Non-normal distributions can also be characterized by the 3rd and 4th moments (skew and kurtosis, respectively), with investors generally preferring higher/more positive skew and lower/less kurtosis. Two portfolios with the same mean and variance can differ in their skew and kurtosis, with diversification playing a role in improving those additional metrics thus making the portfolio more desirable.

These two points are also related to each other (skew and kurtosis can play a role in affecting the relationship between the three Pythagorean means).

I think the most important aspect of what I just mention is the fact that for portfolios in withdrawal (i.e. in retirement) the harmonic mean is very important since it directly relates to the sustainable withdrawal rate (or SWR). I'm not a mathematician, so I'm not sure how you'd set about proving this formally, but I've observed over many simulations that higher levels of portfolio diversification seem to - holding arithmetic mean and variance constant - increase the harmonic mean so that it is closer to the other two.

This is very important for retirees, but not quite so important for accumulators.
"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: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

MichCPA wrote: Tue May 21, 2019 4:23 pm Its a debate about whether a low correlation to stocks makes LT bonds worth owning.
Since I wrote the original post, I feel I have some claim to clarify what the post is "about". The goal of the post was to illustrate which ETFs, and to some degree which categories of ETFs, were most effective at diversifying a portfolio that otherwise consists entirely of US stocks.

Long-term US treasury bond ETFs are indisputably superior to the other Vanguard ETFs in that regard, and not ONLY because of their correlation but also because of their variance.

Whether that makes them "worth owning" depends on other factors (namely the investor's investment horizon along with the desired return and variance characteristics of the portfolio).
MichCPA wrote: Tue May 21, 2019 4:23 pmI would furthermore point out that correlation doesn't by itself say much about risk.
Correlation says NOTHING about the riskiness of an asset in isolation, but it says a lot about the role that asset will play in the riskiness of the portfolio.
MichCPA wrote: Tue May 21, 2019 4:23 pm Interest rates seem to me to be a far preferable lens with which to view bonds, because they are a causative factor, not a correlation which can be muddied by all sorts of variables known and unknown.
I suspect you are not alone in taking this view. One goal of this post (and others like it) is to alert readers that there are more much more modern approaches to portfolio construction than this available, should people want to employ them.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
acegolfer
Posts: 3029
Joined: Tue Aug 25, 2009 9:40 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by acegolfer »

vineviz wrote: Tue May 21, 2019 4:45 pm Two portfolios with the same mean and variance can differ in their skew and kurtosis, with diversification playing a role in improving those additional metrics thus making the portfolio more desirable.
If 2 portfolios have the same E(r), same stdev but 2 different skewness/kurtosis, then they don't have the same risks (unless one thinks stdev is the only risk measurement).
MichCPA
Posts: 884
Joined: Fri Jul 06, 2018 9:06 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by MichCPA »

acegolfer wrote: Tue May 21, 2019 4:56 pm
vineviz wrote: Tue May 21, 2019 4:45 pm Two portfolios with the same mean and variance can differ in their skew and kurtosis, with diversification playing a role in improving those additional metrics thus making the portfolio more desirable.
If 2 portfolios have the same E(r), same stdev but 2 different skewness/kurtosis, then they don't have the same risks (unless one thinks stdev is the only risk measurement).
+1 Ding Ding !
User avatar
Topic Author
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

acegolfer wrote: Tue May 21, 2019 4:56 pm
vineviz wrote: Tue May 21, 2019 4:45 pm Two portfolios with the same mean and variance can differ in their skew and kurtosis, with diversification playing a role in improving those additional metrics thus making the portfolio more desirable.
If 2 portfolios have the same E(r), same stdev but 2 different skewness/kurtosis, then they don't have the same risks (unless one thinks stdev is the only risk measurement).
I think we probably agree here.

Until we have a coherent (and generally accepted) unifying theory of risk, though, I tend to think that most people will continue understand "expected return" and "expected risk" as referring to mean and variance. That's the common usage, and in a public forum like this I feel its the pragmatic approach. Otherwise we'd be footnoting every sentence.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Thesaints
Posts: 5108
Joined: Tue Jun 20, 2017 12:25 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by Thesaints »

vineviz wrote: Tue May 21, 2019 2:22 pm As for the hypothesis that current yield curve and/or yield structure is predictive of future correlations, the evidence I've seen don't support that.
It's not the yield curve, it is the monetary policies and one should be blind not to see it. Whenever an extra 0.25% change in interest rate is discussed, both bonds and stocks go in the same direction. The amount of money on the market (stocks+bonds) is for a large part set by the size of the Fed's balance sheet.
Normally, money would move between stocks and bonds, but its total amount would not change rapidly and substantially. That's what gives you the partial inverse correlation between stocks and LT treasuries that you have observed.
In 2008 the Fed put another 3.5 Trillions on the market, now decreased to just 3 extra trillions. That clearly inflated both bonds and stocks prices and, in fact, in the past ten years the only decorrelation between stocks and LT treasuries was observable on a short time scale (i.e. small movements). What do you think will happen if and when those 3T are taken away anywhere as rapidly as they appeared ?
Angst
Posts: 2968
Joined: Sat Jun 09, 2007 11:31 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by Angst »

vineviz wrote: Tue May 21, 2019 3:13 pm
Angst wrote: Tue May 21, 2019 1:01 pm Vineviz, I'm disappointed you haven't commented yet on my post (below) from last night. Perhaps it just got past you in the course of things.
Sorry, I did indeed overlook this post.
Angst wrote: Mon May 20, 2019 7:07 pm So when the market crashes, would I really prefer to have VTI paired 50:50 with one of these equity ETF's having correlations like .6 vs. any of the above bond funds having correlations ranging from (.35) to .18? I find that counter-intuitive, to say the least. Is that really what you mean?
Solely from a diversification standpoint, you should be indifferent between two choices that have the same diversification ratio. Now, the data I presented was for simple 50/50 pairs of ETFs and virtually one of us have only two funds in our portfolios so we are operating at some level of simplification.

The diversification ratio is effectively a joint outpour of correlation and variance. Since we're taking VTI as a fixed partner, we can inver that if two other ETFs have the same diversification ratio but different correlations then they must also have different variances. Higher variances produce higher diversification ratios for a give correlation, so we can also infer that if one ETF has the same diversification ratio as another ETF but a lower correlation then the first ETF must have a HIGHER variance. Because markets are roughly mean-variance efficient, we can reasonable assume the the asset with higher variance has a higher expected return.

In other words, we can (I think) reasonably infer that a VTI/ETF pair on the blue line will have higher risk-adjusted return than a VTI/ETF pair having the same diversification ratio that is below the blue line.

That's a mouthful.

Angst wrote: Mon May 20, 2019 7:07 pmWhile some people have since been arguing for ST or IT bonds in certain circumstances over your preference for LT, what attracts my attention with respect to this part of the discussion even more is that based on your graphic in the OP you'd even prefer Vanguard Consumer Staples ETF (VDC) over Vanguard Short-Term Bond ETF (BSV). I'm surprised this isn't being discussed as well.
Me too, though in fairness diversification is just one dimension of portfolio optimization. Many investors (especially the relentlessly conservative ones that a forum like this tends to attract) are much more interested in de-risking their portfolio than in diversifying it. Adding a short-term bond fund will reduce the portfolio volatility quickly, even if it actually isn't diversifying the portfolio very much.
Thank you for your reply above - yes, the first part was a mouthful! But I understood what you said and once again, as I said in my first post, really interesting.

Now as much as I'm fascinated with what you've shown and feel I've got something of a handle on it, I still have to wonder how well we can rely upon the results because of the limited history. For example, regarding those equity ETF's I focused on which had diversification ratios similar to the group of ST bond ETF's but also had higher variance with VTI than the ST bond ETF's, here are their Years of Inception:

Code: Select all

Equity ETF's     (VTI - 2001)
VDC   -   2004
VNQ   -   2004
VNQI  -   2010
VWO   -   2005
VOX   -   2004

ST Bond ETF's
BSV   -   2007
VGSH  -   2009
VCSH  -   2009
VTIP  -   2012
Comparing them to VTI you had from as little as 7 years (VTIP) to 15 years of data to work with. Is that enough to rely on your results for setting one's (albeit hypothetical) grand 2-fund portfolio for... well, lets say at least for the next 10 years or so? (I don't want to divert this into a ST vs LT bond discussion.) Equity and bond markets and macroeconomic factors in general over the last 7-15 years for which we have the ETF data have been behaving quite differently than in the 1970's and 1980's. As interesting as your methodology and results are, can we really make use of them given the limited historical data?
pdavi21
Posts: 1296
Joined: Sat Jan 30, 2016 3:04 pm

Re: Vanguard ETFs, ranked by diversification benefit

Post by pdavi21 »

"Everyone else should pick a bond duration based on their investment horizon and call it a day."
+1

The risk that VTI and EDV drop over 30% at the same time is probably higher than the risk that VTI and BND drop 30% at the same time.
This is because it is nearly impossible that BND will ever drop 30%. The average expected outcome of EDV gaining (let's say 10% for every 30% VTI drops) does not eliminate the very real risk that it loses big-a risk which a diversified shorter duration bond fund would not have.

Maybe this is the "skewness" you all were talking about.
"We spend a great deal of time studying history, which, let's face it, is mostly the history of stupidity." -Stephen Hawking
User avatar
Topic Author
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Vanguard ETFs, ranked by diversification benefit

Post by vineviz »

Thesaints wrote: Tue May 21, 2019 5:20 pm Whenever an extra 0.25% change in interest rate is discussed, both bonds and stocks go in the same direction.
And yet . . . .


Image
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Thesaints
Posts: 5108
Joined: Tue Jun 20, 2017 12:25 am

Re: Vanguard ETFs, ranked by diversification benefit

Post by Thesaints »

vineviz wrote: Tue May 21, 2019 7:43 pm And yet . . . .


Image
Rolling 60-day Correlation
in the past ten years the only decorrelation between stocks and LT treasuries was observable on a short time scale (i.e. small movements).
QED
Post Reply