Backtesting International Bonds (as an aggregate)

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Backtesting International Bonds (as an aggregate)

Post by siamond » Sun Nov 20, 2016 6:57 pm

As I was looking at various data series, I started to ponder about International Bonds (as an aggregate, not per country). There are only a few real-life funds for this, and all with very limited history. Vanguard launched the Total International Bond Index Fund (VTIBX) in 2013. A slightly older one is about treasuries only: iShares International Treasury Bond ETF (IGOV), launched in 2009. Also, iShares Core International Aggregate Bond ETF (IAGG) more directly competes with VTIBX (or the ETF equivalent, BNDX), launched in Nov-15.

Clearly, just a few years of history doesn't tell us much. The index VTIBX follows is equally new (2013): Bloomberg Barclays Global Aggregate ex-USD Float Adjusted RIC Capped Index (USD Hedged). So I started to look for similar indexes with a much longer record, matching the 'hedged' property of VTIBX, and found two which seem relevant:
- JP. Morgan (JPM) GBI Global Ex US TR Hdg USD (Morningstar FOUSA06E2D): starting in 1993
- Citigroup WGBI NonUSD Hedged USD (Morningstar XIUSA000SM): starting in 1984

I compared the two data series, and they seem reasonably in sync, and even match the 2014/15 years of VTIBX reasonably well. So I assembled a 1985+ spliced data series with Citigroup index and VTIBX, subtracting the VTIBX ER to the index years. I actually started in 1986 as 1985 seemed a tad idiosyncratic (TBM returned a whopping 22%!). And then I compared five portfolios:
- 100% US Total-Bonds
- 100% International Bonds
- 75% US, 25% International
- 50% US, 50% International
- 25% US, 75% International

Image

One might find attractive the lower std-deviation displayed by the 75/25 portfolio. The Sharpe ratio is indeed improved.

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Sun Nov 20, 2016 7:05 pm

Now let's put that in context of a more realistic portfolio. I used the classic 60/40 asset allocation, and arbitrarily allocated 40% to US equities and 20% to International. Then divided the 40% bonds in the same proportions as the previous test.

If you see a difference in the graphs below, well, take a close look at the values on the axis, there is really nothing meaningful that I can see.

Sure enough, backtesting bonds over 30 years, including the "bonds' bull market" period, is probably not terribly meaningful for what is in front of us, which is likely to be shaped by the currently extremely low yields and the possibility of rising inflation. But still, it is always interesting to look at historical facts, and in this case, I have to say that my idle curiosity about international bonds drained pretty fast... :wink:

Image

What do you think?

User avatar
abuss368
Posts: 11628
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: Backtesting International Bonds (as an aggregate)

Post by abuss368 » Sun Nov 20, 2016 7:52 pm

Am I understanding the charts correctly? There is small (as in 0.00 - two decimal) differences?
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Sun Nov 20, 2016 7:55 pm

abuss368 wrote:Am I understanding the charts correctly? There is small (as in 0.00 - two decimal) differences?

Yes, you are. That is, for the second chart. The difference is a little bit more meaningful for the first chart, but then it's not a real-life portfolio, just a direct comparison between bond types.

User avatar
nisiprius
Advisory Board
Posts: 34115
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: Backtesting International Bonds (as an aggregate)

Post by nisiprius » Sun Nov 20, 2016 8:09 pm

Just as one more potential data source, what about PIMCO Global Bond Fund (U.S. Dollar-Hedged), PGBIX? Goes back to 1995, is dollar-hedged just like VTIBX, and has closely paralleled VTIBX in the time both have been available.
Last edited by nisiprius on Sun Nov 20, 2016 9:22 pm, 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.

stlutz
Posts: 4011
Joined: Fri Jan 02, 2009 1:08 am

Re: Backtesting International Bonds (as an aggregate)

Post by stlutz » Sun Nov 20, 2016 9:18 pm

Nothing particular to add, but all of your work in digging up and assembling better data series for the spreadsheet has been pretty amazing. Thanks to you (and longinvest) for all of your efforts! :sharebeer

AlohaJoe
Posts: 2546
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Backtesting International Bonds (as an aggregate)

Post by AlohaJoe » Sun Nov 20, 2016 9:57 pm

The only time I can remember seeing anyone look at international bonds was Jared Kizer in his 2005 article "Drawing Down and Looking Abroad". It was based on the DMS dataset and he was more interested in a globally diversified portfolio rather than just bonds per se, so it's not hugely useful.

He did find that a globally diversified portfolio beat the US-only portfolio at nearly all asset allocations by a substantial amount. He points out that Cooley et al (2003) also looked at globally diversified portfolios and found it didn't help; but their analysis was based on Monte Carlo rather than historical backtesting and they only included global equities not global bonds.

So there is a slight (very slight!) implication that global bonds may have helped. But he doesn't really show a with/without to measure the impact. All we really know is the raw data provided by DMS about returns and standard deviation: US-only bonds returned 5.2% with 8.2% standard deviation while a globally-diversified set of bonds returned 5.5% with 7.9% standard deviation (based on data from 1900 to 2003).

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Sun Nov 20, 2016 11:18 pm

nisiprius wrote:Just as one more potential data source, what about PIMCO Global Bond Fund (U.S. Dollar-Hedged), PGBIX? Goes back to 1995, is dollar-hedged just like VTIBX, and has closely paralleled VTIBX in the time both have been available.

We already have something close to that in the Simba spreadsheet, with the global bonds (US included) unhedged data series, which is essentially the following spliced series:
* JP Morgan Global Govt Bond (unhedged) 1987-1996
--> which could be replaced by Citigroup WGBI USD (unhedged) to start in 1985
* Pimco Global Bond Fund (Unhedged): PIGLX 1994+

Now, what you're referring to is the hedged variant of the same thing, and it could be something like:
* Citigroup WGBI Hedged USD (Morningstar XIUSA000T5) 1985-1995
* Pimco Global Bond Fund (U.S. Dollar-Hedged): PGBIX 1996+

I ran a quick test just with the Citigroup hedged series and the VTIBX ER, and as you hinted at, it didn't make any meaningful difference. I am all for international exposure, but I'll stick to the equity side of things!

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Sun Nov 20, 2016 11:26 pm

stlutz wrote:Nothing particular to add, but all of your work in digging up and assembling better data series for the spreadsheet has been pretty amazing. Thanks to you (and longinvest) for all of your efforts!

Welcome, your words of appreciation are very encouraging. I'm getting pretty close to completion on the major refresh I've been working on, this was quite the discovery process!

In this precise case though, although I would really want to add such an Int'l Bonds (hedged) data series to the Simba spreadsheet, so that people can easily redo the analysis I just did or some variant of it, I am not sure this would be wise. Although I found all the numbers on Morningstar, and this is all for personal education purposes, I know that Citigroup is a little tight about the idea of sharing their indices numbers... :|

User avatar
abuss368
Posts: 11628
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: Backtesting International Bonds (as an aggregate)

Post by abuss368 » Tue Nov 22, 2016 9:14 am

siamond wrote:
abuss368 wrote:Am I understanding the charts correctly? There is small (as in 0.00 - two decimal) differences?

Yes, you are. That is, for the second chart. The difference is a little bit more meaningful for the first chart, but then it's not a real-life portfolio, just a direct comparison between bond types.


Hi siamond,

Do you invest in international bonds or specifically the Vanguard Total International Bond Index Fund?

Best.
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Tue Nov 22, 2016 9:50 am

abuss368 wrote:Do you invest in international bonds or specifically the Vanguard Total International Bond Index Fund?

I do not. Never quite saw the point. And I have to say that my little backtesting exercise didn't exactly encourage me to change my mind... :wink:

User avatar
abuss368
Posts: 11628
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: Backtesting International Bonds (as an aggregate)

Post by abuss368 » Tue Nov 22, 2016 9:56 am

siamond wrote:
abuss368 wrote:Do you invest in international bonds or specifically the Vanguard Total International Bond Index Fund?

I do not. Never quite saw the point. And I have to say that my little backtesting exercise didn't exactly encourage me to change my mind... :wink:


Thank you. That is where I am at. I have learned much about these bonds and asset class although much more remains. However, I have not quite made a decision to invest capital.
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

User avatar
nisiprius
Advisory Board
Posts: 34115
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: Backtesting International Bonds (as an aggregate)

Post by nisiprius » Tue Nov 22, 2016 12:30 pm

It was Vanguard's own paper, Global fixed income: Considerations for U.S. investors that convinced me not to bother. But I'm pleased to see your confirmation. The evidence for any important different from international stock investing is thin. The evidence for international bond investing is thinner. I can't bring myself to believe that Vanguard's use of this fund in their all-in-one products does any harm or any good. I'd love to know why Vanguard has been pushing it so hard. I keep thinking that Vanguard believes it must internationalize for it's own good.
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
abuss368
Posts: 11628
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: Backtesting International Bonds (as an aggregate)

Post by abuss368 » Tue Nov 22, 2016 1:07 pm

nisiprius wrote:It was Vanguard's own paper, Global fixed income: Considerations for U.S. investors that convinced me not to bother. But I'm pleased to see your confirmation. The evidence for any important different from international stock investing is thin. The evidence for international bond investing is thinner. I can't bring myself to believe that Vanguard's use of this fund in their all-in-one products does any harm or any good. I'd love to know why Vanguard has been pushing it so hard. I keep thinking that Vanguard believes it must internationalize for it's own good.


I believe this could be the case as well. Oh we have the stock funds for both U.S. and International. We have REIT Funds for both U.S. and International. Heck we have just about all funds for both. Why not bonds to round everything out?

Best.
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Tue Nov 22, 2016 1:32 pm

abuss368 wrote:Oh we have the stock funds for both U.S. and International. We have REIT Funds for both U.S. and International. Heck we have just about all funds for both. Why not bonds to round everything out?

Because bonds are not stocks? Stock diversification is definitely valuable, given the erratic behavior of equities. I just don't see the rationale (nor the historical justification) for bonds diversification (I mean, beyond domestic borders). Different goals apply to stocks and bonds.

Beliavsky
Posts: 688
Joined: Sun Jun 29, 2014 10:21 am

Re: Backtesting International Bonds (as an aggregate)

Post by Beliavsky » Tue Nov 22, 2016 2:24 pm

Government bonds in several countries have negative yields. (At least they did a month ago. Since then rates have risen.) I would be interested in investing in a low-cost semi-active international bond fund that invests in bond markets around the world but which avoids bonds with negative yields.

Beliavsky
Posts: 688
Joined: Sun Jun 29, 2014 10:21 am

Re: Backtesting International Bonds (as an aggregate)

Post by Beliavsky » Tue Nov 22, 2016 2:27 pm

siamond wrote:
abuss368 wrote:Oh we have the stock funds for both U.S. and International. We have REIT Funds for both U.S. and International. Heck we have just about all funds for both. Why not bonds to round everything out?

Because bonds are not stocks? Stock diversification is definitely valuable, given the erratic behavior of equities. I just don't see the rationale (nor the historical justification) for bonds diversification (I mean, beyond domestic borders). Different goals apply to stocks and bonds.

The rationale for currency-hedged international bond investment is that excess returns of bonds over cash around the world are less than 100% correlated. If there is a term premium in Canada, the UK, Germany, Japan, Australia etc. that is distinct from the U.S. term premium, why not get exposure to it?

AlohaJoe
Posts: 2546
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Backtesting International Bonds (as an aggregate)

Post by AlohaJoe » Tue Nov 22, 2016 9:45 pm

nisiprius wrote:The evidence for any important different from international stock investing is thin. The evidence for international bond investing is thinner. I can't bring myself to believe that Vanguard's use of this fund in their all-in-one products does any harm or any good. I'd love to know why Vanguard has been pushing it so hard.


I don't think the reason is so mysterious. It only looks that way if you start out with default assumption of 100% home bias and insist on evidence for why you should invest elsewhere.

If you start out with the assumption that markets are efficient and a global cap weighting makes sense, then that same "thin evidence" is easily seen as "thin evidence for home country bias".

What siamond sees as "no compelling benefit" I see as "no evidence of downside".

If someone doesn't have an existing portfolio with the inertia (and possible capital gains) that go with that, then I don't see any compelling reason not to direct new money to international bonds other than irrational home country bias, since there is no evidence it hurts.

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Tue Nov 22, 2016 11:36 pm

AlohaJoe wrote:If someone doesn't have an existing portfolio with the inertia (and possible capital gains) that go with that, then I don't see any compelling reason not to direct new money to international bonds other than irrational home country bias, since there is no evidence it hurts.

You have a good point, and I am as wary as you are about home country bias, and I do like to hedge my bets. My equities are 50/50 domestic/international, as a case in point. Same for my REITs. But I don't want to make my portfolio more complicated just for the sake of it either.

It seems to me that you're making a powerful case for a Global Bonds index fund (hedged in one's domestic currency). Something close to what nisiprius mentioned, actually, e.g. PIMCO Global Bond Fund (U.S. Dollar-Hedged), PGBIX? But truly passive, and with a low ER (not 0.55% - and that's institutional!). Trouble is I don't believe there is such a fund from the major fund providers. There is an index for it though: Bloomberg Barclays Global Aggregate (USD hedged) Index... Personally, I would probably go for something like that, to replace my US bonds, not to add to them.

AlohaJoe
Posts: 2546
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Backtesting International Bonds (as an aggregate)

Post by AlohaJoe » Wed Nov 23, 2016 8:55 am

siamond wrote:Something close to what nisiprius mentioned, actually, e.g. PIMCO Global Bond Fund (U.S. Dollar-Hedged), PGBIX? But truly passive, and with a low ER (not 0.55% - and that's institutional!). Trouble is I don't believe there is such a fund from the major fund providers. There is an index for it though: Bloomberg Barclays Global Aggregate (USD hedged) Index... Personally, I would probably go for something like that, to replace my US bonds, not to add to them.


I overstated my case; there is definitely value in simplicity. If you're investing in a Target Date Fund and Vanguard or Fidelity are handling things, then I think it is a no-brainer for them to add in International Bonds. But if you're managing your own portfolio, I can understand not wanting to add another fund. Everyone has to draw the line somewhere!

I'm not aware of any decent all-in-one bond funds either. Lots of ex-US funds but nothing that combines them all. http://etfdb.com/index/barclays-capital ... ond-index/ seems to back that up: there's nothing (that they know of, at least) that tracks the Barclays Global Aggregate Index.

User avatar
abuss368
Posts: 11628
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: Backtesting International Bonds (as an aggregate)

Post by abuss368 » Wed Nov 23, 2016 8:57 am

I am curious if the Federal TSP will ever include international bonds. If that becomes reality, in my opinion, that is a huge feather in the cap for this asset class.
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

columbia
Posts: 504
Joined: Tue Aug 27, 2013 5:30 am

Re: Backtesting International Bonds (as an aggregate)

Post by columbia » Fri Nov 25, 2016 3:31 pm

I see that there is about $64B in the VG international bond index. At what level of investment would it take for the fund to have any tangible effect (i.e. lower demand) on the market for US bonds?

AlohaJoe
Posts: 2546
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Backtesting International Bonds (as an aggregate)

Post by AlohaJoe » Fri Nov 25, 2016 5:25 pm

columbia wrote:I see that there is about $64B in the VG international bond index. At what level of investment would it take for the fund to have any tangible effect (i.e. lower demand) on the market for US bonds?


Probably around $5 trillion or so.

User avatar
abuss368
Posts: 11628
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: Backtesting International Bonds (as an aggregate)

Post by abuss368 » Fri Nov 25, 2016 8:53 pm

columbia wrote:I see that there is about $64B in the VG international bond index. At what level of investment would it take for the fund to have any tangible effect (i.e. lower demand) on the market for US bonds?


What is the total value of Total Bond Index fund?
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Mon Nov 28, 2016 10:36 pm

abuss368 wrote:What is the total value of Total Bond Index fund?

If you mean the US fund (e.g. VBMFX), it is currently $175B. That's a big number! Although not that oversized in comparison to the Int'l fund (VTIBX, at $64B, as previously mentioned). Hm. Quite a few investors do like those Int'l bonds... Not too sure why, but they do.

BetaTracker
Posts: 158
Joined: Fri Sep 21, 2012 7:57 am

Re: Backtesting International Bonds (as an aggregate)

Post by BetaTracker » Mon Nov 28, 2016 11:25 pm

What about portfolio visualizer? It has a selection under the asset allocation backtesting tool for USD Hedged global bonds (and unhedged) -- goes back to 1999. It shows that in 2008 this asset class lost -2.35%.
Personally, I face a quandary since talking to a Vanguard advisor recently who informed me that PAS is no longer accepting people who aren't willing to allocate at least 20% to international bonds. (Perhaps I just heard wrong and he was talking about himself?) In any case, I was counting on PAS as our backup plan for my wife since she has no interest in portfolio management.
I'd rather be content than happy -- Lao Tzu.

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Mon Nov 28, 2016 11:30 pm

Let's take a look at the choice between unhedged and hedged international bonds now. I used the following data series:
- Citigroup WGBI NonUSD USD (i.e. Int'l government bonds, unhedged)
- Citigroup WGBI NonUSD Hedged USD (i.e. Int'l government bonds, hedged)
- Citigroup US GBI USD (i.e. US government bonds)

I used all of those index data series with the same ER of 0.20 (e.g. close to what Vanguard uses for various bond funds), to make it more realistic and yet compare apple to apple. All those data series start in 1985, and afaik, those are the longest ones (JP Morgan and Barclays similar series start later). I compared the following portfolios with various doses of (US, Int'l unhedged, Int'l hedged) bonds:
- Bd 100/0/0 (all US)
- Bd 50/50/0 (mix)
- Bd 50/0/50 (mix)
- Bd 50/25/25 (mix)
- Bd 0/0/100 (all Int'l)

Image

Based on such all-bonds trajectories, one might be tempted to diversify.

Looking at the telltale chart, there is some steep difference to begin with, this was the tail end of the oil crisis, and Int'l Bonds (unhedged) grew 30% a year for 3 years in a row, holy smokes, but then of course, the roller-coaster went crashing down. Personally, Int'l Bonds (unhedged) look way too much like equities for my taste!

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Mon Nov 28, 2016 11:38 pm

Now let's look at a more realistic portfolio. I used 25% US equities, 25% Int'l equities, and the rest is bonds, in the same proportions as the previous post. Here is the trajectory. Again, pay attention to the telltale after the somersaults of the late 80s.

Image

The differences in return are puny, while the differences in volatility are more significant. And... I just don't see the case for international bond diversification any more (at least in past data, future might be different of course). This being said, it is too bad that the data series do not go back to 1970 or so, as it would have been really interesting to see what happened during the oil crisis and its aftermath.

EDIT: forgot to mention that Sharpe and Sortino ratios both favor the US-bonds-only portfolio in those tests including equities.
Last edited by siamond on Tue Nov 29, 2016 12:36 am, edited 1 time in total.

User avatar
abuss368
Posts: 11628
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: Backtesting International Bonds (as an aggregate)

Post by abuss368 » Mon Nov 28, 2016 11:58 pm

BetaTracker wrote:What about portfolio visualizer? It has a selection under the asset allocation backtesting tool for USD Hedged global bonds (and unhedged) -- goes back to 1999. It shows that in 2008 this asset class lost -2.35%.
Personally, I face a quandary since talking to a Vanguard advisor recently who informed me that PAS is no longer accepting people who aren't willing to allocate at least 20% to international bonds. (Perhaps I just heard wrong and he was talking about himself?) In any case, I was counting on PAS as our backup plan for my wife since she has no interest in portfolio management.


Wow! Really! Vanguard requires an investment in international bonds to be a client?
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

lack_ey
Posts: 5690
Joined: Wed Nov 19, 2014 11:55 pm

Re: Backtesting International Bonds (as an aggregate)

Post by lack_ey » Tue Nov 29, 2016 12:05 am

Did you account for or otherwise know how to estimate hedging costs (that would be incurred by a fund relative to the index), when looking at the hedged data series?

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Tue Nov 29, 2016 12:15 am

lack_ey wrote:Did you account for or otherwise know how to estimate hedging costs (that would be incurred by a fund relative to the index), when looking at the hedged data series?

I used indices which take in account hedging in the index itself. And as you can see, the data series are quite different from the unhedged ones (and when running a quick comparison between Citi, JP Morgan and Barclays, they are reasonably in sync for a given type - hedged or unhedged).

I have no basis for estimating additional costs and tracking errors from real-life funds, as funds like VTIBX are very new. I would just hope that Vanguard would be true to its excellent track record in this respect. Do you have any reason to doubt it?

lack_ey
Posts: 5690
Joined: Wed Nov 19, 2014 11:55 pm

Re: Backtesting International Bonds (as an aggregate)

Post by lack_ey » Tue Nov 29, 2016 12:42 am

siamond wrote:
lack_ey wrote:Did you account for or otherwise know how to estimate hedging costs (that would be incurred by a fund relative to the index), when looking at the hedged data series?

I used indices which take in account hedging in the index itself. And as you can see, the data series are quite different from the unhedged ones (and when running a quick comparison between Citi, JP Morgan and Barclays, they are reasonably in sync for a given type - hedged or unhedged).

I have no basis for estimating additional costs and tracking errors from real-life funds, as funds like VTIBX are very new. I would just hope that Vanguard would be true to its excellent track record in this respect. Do you have any reason to doubt it?

The hedging is not frictionless. There's a bid-ask spread for the currency forwards that I presume is eating into a fund return but probably not accounted for in the index return.

I think somewhere Vanguard quoted a 20 bp figure for cost but that was an estimate before they started running the fund I think. I haven't checked to verify that.

neosmile
Posts: 2
Joined: Sun Feb 12, 2017 7:49 pm

Re: Backtesting International Bonds (as an aggregate)

Post by neosmile » Sun Feb 12, 2017 7:51 pm

Siamond, may I ask how one can download the price histories of the following you mentioned for backtesting purposes? I can't find any sources on the web...

- JP. Morgan (JPM) GBI Global Ex US TR Hdg USD (Morningstar FOUSA06E2D): starting in 1993
- Citigroup WGBI NonUSD Hedged USD (Morningstar XIUSA000SM): starting in 1984

Thanks in advance!

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Sun Feb 12, 2017 11:46 pm

neosmile wrote:Siamond, may I ask how one can download the price histories of the following you mentioned for backtesting purposes? I can't find any sources on the web...

- JP. Morgan (JPM) GBI Global Ex US TR Hdg USD (Morningstar FOUSA06E2D): starting in 1993
- Citigroup WGBI NonUSD Hedged USD (Morningstar XIUSA000SM): starting in 1984

Thanks in advance!

Here are two magical URLs, derived from the Morningstar SecId I provided. This requires a bit more massaging, but it's pretty easy to do with a text import in Excel, and some subsequent clean-up. I really hope that Morningstar is not going to screw that up with their ongoing Web site update... :?

http://quotes.morningstar.com/indexquote/quote.html?t=FOUSA06E2D

http://mschart.morningstar.com/chartweb/defaultChart?type=getcc&secids=FOUSA06E2D&dataid=8225&startdate=1900-01-01&enddate=2019-01-05&currency=&format=1

neosmile
Posts: 2
Joined: Sun Feb 12, 2017 7:49 pm

Re: Backtesting International Bonds (as an aggregate)

Post by neosmile » Mon Feb 13, 2017 7:30 pm

Thanks so much, this is very helpful!

Dan$
Posts: 1
Joined: Wed Jun 07, 2017 4:51 pm

Re: Backtesting International Bonds (as an aggregate)

Post by Dan$ » Wed Jun 07, 2017 5:00 pm

Hi Siamond,
I am just coming to this exchange for the first time and find it very helpful.

So what you are basically saying is in order to look up what the historical record might be of Vanguard's Total International Bond Index Fund, you gathered data that matched the funds benchmark: Bloomberg Barclays Global Aggregate ex-USD Float Adjusted RIC Capped Index (USD Hedged).

And what you fund is that over the past few years the returns have not been so great.

Is that how you would put it in simple terms?
Dan

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Wed Jun 07, 2017 9:18 pm

Dan$ wrote:Hi Siamond,
I am just coming to this exchange for the first time and find it very helpful.
Cool.
Dan$ wrote:So what you are basically saying is in order to look up what the historical record might be of Vanguard's Total International Bond Index Fund, you gathered data that matched the funds benchmark: Bloomberg Barclays Global Aggregate ex-USD Float Adjusted RIC Capped Index (USD Hedged).
Yes. We actually use such technique quite often in other backtesting endeavors. This seems to make solid sense, given that the fund is supposed to track its benchmark (and Vanguard proved that they are pretty good at it with other -older- bond funds).
Dan$ wrote:And what you found is that over the past few years the returns have not been so great. Is that how you would put it in simple terms?
Well, not exactly. What I found is that over the past 30 years, there would have been no observable benefit in diversifying one's bonds between domestic (US) and International for a realistic Asset Allocation (i.e. including some amount of stocks). I'm not saying that Int'l Bonds are bad per se, I don't think they are, actually, but I just don't see the historical evidence for adding those to one's portfolio. This being said, they didn't seem to hurt either, so I have no beef with the fact that Vanguard uses them in LifeStrategy or TargetRetirement funds (of funds).

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: Backtesting International Bonds (as an aggregate)

Post by NiceUnparticularMan » Thu Jun 08, 2017 6:13 am

Also late to this party, but a quick comment on Vanguard:

Part of what they are arguing is international bonds can help reduce volatility a bit more in high-fixed-income portfolios, and that they would have done so historically. That doesn't really suggest people with only moderate or low fixed-income percentages need them, but they might not hurt much either.

But they are also suggesting their forward-looking asset model is predicting a bit more advantage to holding both international equities and international bonds in the future, which is why they upped the percentages in their Target/LifeStrategy funds to 40% and 30% respectively in early 2015.

That second sort of claim is going to be outside the scope of historic analysis.

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Thu Jun 08, 2017 7:51 am

NiceUnparticularMan wrote:[...] they are also suggesting their forward-looking asset model is predicting a bit more advantage to holding both international equities and international bonds in the future, which is why they upped the percentages in their Target/LifeStrategy funds to 40% and 30% respectively in early 2015.

That second sort of claim is going to be outside the scope of historic analysis.
Actually, let's look at another historical perspective. We know that a decade or two of rising interest rates are typically not great for bond returns. And it seems reasonable to expect that the trajectory of the interest rates in the US will be going either upwards or sideways for a while. So one might indeed seek international diversification hoping for a slightly better scenario. Now here is a possible counterpoint. I assembled the following chart for another study, it shows the trajectory of government bonds interest rates (~10 years) in various countries (identified by their respective currencies) since 1970. Click on the image to see a bigger version.

Image

There was undoubtedly a good deal of diversity until the mid 90s. Since then, aside from Switzerland (CHE) and Japan (JPN), there is a clearly display of 'pack mentality'. Now is that something specific to the past 20 years that will not continue in the coming 20 years, I don't know. But at the intuitive level, it makes sense that the various central banks communicate more and more, and then tend to sync up. And the existence of the Euro (since 1999) reinforces the point. The world is flat(ter).

Personally, I do believe that, as a matter of principle, maximum diversification should be the starting point, and see no harm in doing so, Vanguard's strategy is therefore fine with me. But I keep not seeing a strong case for individual investors to add International Bonds. I wish Vanguard would open a global bonds fund though, as I might go for that (diversifying without complexifying, that would be good).

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: Backtesting International Bonds (as an aggregate)

Post by NiceUnparticularMan » Thu Jun 08, 2017 11:08 am

siamond wrote:Personally, I do believe that, as a matter of principle, maximum diversification should be the starting point, and see no harm in doing so, Vanguard's strategy is therefore fine with me. But I keep not seeing a strong case for individual investors to add International Bonds.
So here is the problem. Vanguard's forward-looking asset model is proprietary. They are willing to describe it in general terms. They are also willing to describe in general terms how they use it to set their asset allocations in Target funds, and when using it to give advice to investors. Here is such a paper:

https://personal.vanguard.com/pdf/ISGGCM.pdf

And here is a nice summary figure:

Image

However, to my knowledge they don't release the details of what they are doing to the public, nor the detailed results, beyond these pretty general white papers.

So yeah, you probably haven't seen their actual case in any detail, and neither have I. That's because they don't want to tell us.

User avatar
zonto
Posts: 75
Joined: Tue Aug 24, 2010 3:45 pm
Location: Boston, MA

Re: Backtesting International Bonds (as an aggregate)

Post by zonto » Tue Nov 14, 2017 3:23 pm

I've been researching international bond funds lately, and found the Bogleheads forum to have a general distaste for them. I find this somewhat odd, especially given the forum's relative acceptance of bond funds like Vanguard Intermediate-Term Bond Index (VBILX), which is comprised of 50% intermediate-term Treasuries and 50% intermediate-term corporate bonds. VBILX is the best intermediate bond fund I have access to in my 401(k), but I wanted to see if I could potentially improve upon it via funds available through Fidelity's BrokerageLink service. I think I can.

Below are backtest data from Portfolio Visualizer from Jan. 2003 through Sept. 2017 (annually rebalanced, as applicable).

Legend for selected entries:
  • TBM = total U.S. bond market
  • ITT = intermediate-term Treasury bonds
  • Corp = corporate bonds
  • GB(H) = global bonds (USD hedged)
  • TIPS = U.S. Treasury inflation-protected securities

Code: Select all

					CAGR	Stdev	Sharpe		Sortino		U.S. Market Correlation
Individual Assets:
100% TBM				4.05%	3.45%	0.83		1.35		-0.02
100% ITT				3.89%	4.56%	0.61		1.01		-0.22
100% GB(H)				5.42%	3.70%	1.11		1.86		+0.15
100% TIPS				4.19%	6.12%	0.51		0.75		+0.13
100% Corp				5.42%	7.37%	0.59		0.95		+0.25

Blended Portfolios:
50% ITT / 50% Corp*			4.69%	5.31%	0.67		1.10		+0.08
50% ITT / 50% GB(H)			4.71%	3.67%	0.95		1.64		-0.06
70% TBM / 30% GB(H)**			4.47%	3.33%	0.97		1.61		+0.03
50% ITT / 50% TIPS***			4.09%	4.97%	0.60		0.94		-0.02
50% ITT / 30% GB(H) / 20% TIPS****	4.46%	4.10%	0.80		1.32		-0.04
(source) (only allows three portfolios to be tested concurrently)

* Basically equivalent to VBILX.
** Vanguard's current bond allocation in LifeStrategy and Target Date funds.
*** David Swenson's recommended bond allocation.
**** "zonto's amalgam" using Swenson's suggested 50% ITT allocation, Vanguard's suggested 30% GB(H) allocation, and Rick Ferri's suggested 20% TIPS allocation.


Observations:
  • In this test, replacing the 50% corporate bond slice of VBILX with GB(H) resulted in nearly identical CAGR to VBILX, but with a 31% reduction in volatility. This is because GB(H) offered the highest return and risk-adjusted return over the period.
  • Vanguard's suggested 70% TBM / 30% GB(H) bond allocation reduced volatility and increased CAGR compared to 100% TBM. It also offered attractive risk-adjusted returns compared to anything else tested above.
  • Treasuries' inverse correlation with the U.S. stock market make them attractive, especially for those like me with higher equity allocations. I know many on the forum use ITT instead of TBM.
----------

I get the appeal of the total U.S. bond market index. However, when I examine the risk-adjusted performance and positive correlation to the U.S. market of the corporate bonds held within it, I think it makes more sense, if one is going to diversify away from Treasuries, to do so with global bonds (USD hedged) instead. Yes, foreign governments may not be as credit-worthy as the U.S. government, but aren't they significantly more credit-worthy than U.S. companies? Agency bond ratings seem to indicate so. Not only that, but corporate bonds exhibited by far the highest correlation to the U.S. stock market, exactly what an investor doesn't want in a time of crisis. Portfolio Visualizer did not permit backtesting of mortgage-backed securities.

I'm curious as to others' thoughts here.

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Tue Nov 14, 2017 4:20 pm

zonto wrote:
Tue Nov 14, 2017 3:23 pm
I've been researching international bond funds lately.
[...]
Below are backtest data from Portfolio Visualizer from Jan. 2003 through Sept. 2017 (annually rebalanced, as applicable).
[...]
I'm curious as to others' thoughts here.
Hi zonto. Thank you for sharing your findings. I would caution you a bit about reaching conclusions from a specific (and rather short) period of time, although unfortunately, we just don't have that much history about International Bonds to assess.

I wrote the original post roughly one year ago, as my first rough attempt to grasp this issue. Since then, I did a more thorough study as part of a blog article that you can find by clicking here. And I started to realize that there is probably more to International bonds than I originally thought. Still, the case isn't fully convincing and I concluded by saying:

International Bonds (at least in a currency hedged flavor) did not seem to add much (nor hurt in any way). If Vanguard would offer a low cost Global Bond fund, it would probably be reasonable to go for it, but splitting US Bonds to create room for International Bonds doesn’t seem terribly useful.

Anyhoo, you will probably be interested by this little study. Comments welcome.

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Tue Nov 14, 2017 4:30 pm

zonto wrote:
Tue Nov 14, 2017 3:23 pm
I find this somewhat odd, especially given the forum's relative acceptance of bond funds like Vanguard Intermediate-Term Bond Index (VBILX), which is comprised of 50% intermediate-term Treasuries and 50% intermediate-term corporate bonds.
[...]
I get the appeal of the total U.S. bond market index. However, when I examine the risk-adjusted performance and positive correlation to the U.S. market of the corporate bonds held within it, I think it makes more sense, if one is going to diversify away from Treasuries, to do so with global bonds (USD hedged) instead.
I am not a strong believer in bonds' safety given their abysmal track record during or after wars, but treasuries clearly displayed the lowest correlation with stocks in times of crisis. So I probably should amend my blog conclusion by saying that I'd be glad to have access to a Global Treasury Bonds Fund. In other words, I see where you are coming from.

I actually am a VBILX investor. Your line of thinking is intriguing. Let me dig up the spreadsheet I used for my blog study, and see if I can add a few tests following this line of reasoning...

User avatar
zonto
Posts: 75
Joined: Tue Aug 24, 2010 3:45 pm
Location: Boston, MA

Re: Backtesting International Bonds (as an aggregate)

Post by zonto » Tue Nov 14, 2017 4:45 pm

siamond wrote:
Tue Nov 14, 2017 4:20 pm
. . . Anyhoo, you will probably be interested by this little study. Comments welcome.
Thanks for the reminder. I recall we spoke about the study in the international REIT thread, but I forgot its breadth and inclusion of international bonds. The Portfolio Visualizer data set was limited to 2003 due to their corporate bond data. TIPS go back to 2001 and global bonds (USD hedged) to 1999 I believe.

This quote from your study also stood out to me: "This outcome is somewhat surprising though from the perspective of the negative correlations displayed by Int’l Bonds in times of crisis. This shows that correlations are one thing, but the dynamics of absolute returns at the portfolio level are another. We often forget that correlation is about directionality, but doesn’t capture the amplitude of the change."

What are your thoughts on international bonds vs. corporate bonds and mortgage-backed securities (i.e., using Treasuries + international bonds in lieu of TBM which also includes corporate (~27% currently) and even mortgage-backed securities (~23% currently))? One thing I don't like about the Bloomberg Barclays US Aggregate Bond Index is its ever-fluctuating allocations to corporate bonds and mortgage/asset-backed securities.

EDIT: Just saw your second post above.

I searched a bit for global treasury funds and found the following: Best options are still probably:
  • BNDX/VTABX - Vanguard's total international bond index (USD-hedged, float-adjusted, but only caps individual issuers at 20% so Japan is 1/5 of the portfolio).
  • IAGG - iShares Core International Aggregate Bond ETF - USD-hedged, not float-adjusted, but caps individual issuers (in theory) at 10%. Currently, France is 13.57% and Japan is 12.81% of the portfolio. Lower expense ratio than BNDX, but a higher bid/ask spread and less liquidity. 66% in international treasury bonds.

User avatar
siamond
Posts: 3422
Joined: Mon May 28, 2012 5:50 am

Re: Backtesting International Bonds (as an aggregate)

Post by siamond » Tue Nov 14, 2017 5:46 pm

zonto wrote:
Tue Nov 14, 2017 4:45 pm
What are your thoughts on international bonds vs. corporate bonds and mortgage-backed securities (i.e., using Treasuries + international bonds in lieu of TBM which also includes corporate (~27% currently) and even mortgage-backed securities (~23% currently))?
Your line of reasoning really intrigued me, so I gave it a quick try. Turns out that I already have a proper data series in my monthly spreadsheet. VBILX tracks the Bloomberg Barclays U.S. 5–10 Year Government/Credit Bond Index, which has solid history back to January 1976, cool.

First, I checked the correlations in times of crisis (Internet and Financial), and here you are absolutely right, treasuries (US or International) look much better than the VBILX-equivalent (50% US Treasuries and 50% US Corporates). Which certainly makes sense.

Then I did the same thing as for my blog study, I defined two 60/40 portfolios (30% US, 30% Int'l, 40% Bonds), and I tried two allocations for the bonds (40% VBILX-like; 20% IT Treasuries, 20% Int'l Bonds Hedged) and looked at the 1985-2016 period. Well, the outcome wasn't quite convincing:
- the growth chart is better with VBILX (CAGR 9.4% against 9.1%): corporate bonds do provide better rewards for slight add'l risk
- the drawdown chart is NOT improved for the major crises (plus the 2nd strategy presents some annoying drops in 1987 and 1990)

Image

Hm. That's too bad. I really had sympathy for the idea. But, unless I screwed up my math somewhere, the historical record doesn't seem to make a proper case for it. A similar test using the VBILX index against 40% IT Treasuries (US only) displayed fairly similar results, including the hiccups in 1987/90). VBILX looks pretty good to me!

PS. I don't have solid data for mortgage-based securities. I tried at one point, didn't find enough history if I remember well. Note that VBILX doesn't include any.

Post Reply