Correlations among different types of bonds

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
Topic Author
grabiner
Advisory Board
Posts: 35307
Joined: Tue Feb 20, 2007 10:58 pm
Location: Columbia, MD

Correlations among different types of bonds

Post by grabiner »

Morningstar article: Should You Diversify Your Bond Exposure?

The article includes a chart of correlations among different segments of the bond market, over the 15-year period from 2/1/03-1/31/18. It doesn't get all the segments: the only corporate-only segments are short-term and high-yield. Still, there are some interesting conclusions. (Caution: While this is a long period, it includes only one bear market, 2008; the relationship of stocks to bonds in the next bear market, when you most need diversification, may be different.)

The stock market has a -.3 correlation with Treasury bonds, regardless of duration; that is, there is no additional correlation with interest rates. The correlation with short-term corporate bonds is +.3. That is, Treasuries are significantly better than corporates at diversifying the risk of a stock-heavy portfolio.

Cash (3-6 month Treasuries) has no correlation with anything except for a small positive correlation with short-term Treasuries.

Short-term Treasuries and short-term corporates have only a 0.39% correlation, which suggests a benefit for diversification between different types of bonds.

The Barclays Aggregate Index has an 0.88 correlation with intermediate-term Treasuries. Since the index itself is 42% Treasuries and 21% GNMAs, this still implies a fairly high correlation between intermediate-term Treasuries and intermediate-term corporates. (There is still the same difference in risk behavior, but the interest-rate sensivity affects both.)

Intermediate-term and long-term bonds have very high correlation, 0.93 for aggregate to aggregate, and 0.91 for Treasury to Treasury. The returns of both are dominated by the effect of interest rates; long-term bonds are more sensitive to rates, but both are sensitive at the same time.

High-yield bonds have a 0.72 correlation with stocks, which is more than the correlation with any bond category; they also have an 0.60 correlation with short-term corporate bonds, and a negative correlation with Treasuries of any duration. This does suggest viewing a high-yield bond fund as half stock and half bond. (A fund of half stock and half bonds would have 0.71 correlation with stock and 0.71 correlation with bonds if the stock and bond returns were independent.)
Wiki David Grabiner
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 10:55 pm

Re: Correlations among different types of bonds

Post by lack_ey »

Did it specify anywhere if the correlation was for the daily return series, monthly, or what?
grabiner wrote: Sun Feb 11, 2018 7:54 pmHigh-yield bonds have a 0.72 correlation with stocks, which is more than the correlation with any bond category; they also have an 0.60 correlation with short-term corporate bonds, and a negative correlation with Treasuries of any duration. This does suggest viewing a high-yield bond fund as half stock and half bond.
Why look at correlation and not beta for that?
grabiner wrote: Sun Feb 11, 2018 7:54 pm (A fund of half stock and half bonds would have 0.71 correlation with stock and 0.71 correlation with bonds if the stock and bond returns were independent.)
My thinking cap is not on right now and I may have missed something, but that seems wrong unless maybe you're also assuming equal volatility of stocks and bonds.
livesoft
Posts: 86080
Joined: Thu Mar 01, 2007 7:00 pm

Re: Correlations among different types of bonds

Post by livesoft »

The article doesn't seem to mention short-term corporate bonds, yet you mentioned their correlations. Did you calculate them with portfoliovisualizer.com?

Ah, I think I understand the table labels now.
Wiki This signature message sponsored by sscritic: Learn to fish.
User avatar
happysteward
Posts: 465
Joined: Tue Jun 16, 2015 11:42 am
Location: East Coast USA

Re: Correlations among different types of bonds

Post by happysteward »

Treasuries are significantly better than corporates at diversifying the risk of a stock-heavy portfolio.
Doesn’t this argue for a pure treasury bond fund over and above a core bond fund at least as far as portfolio “ballast” goes?
"How much money is enough?", John Rockefeller responded, "...just a little bit more."
User avatar
patrick013
Posts: 3301
Joined: Mon Jul 13, 2015 7:49 pm

Re: Correlations among different types of bonds

Post by patrick013 »

If we could sin for a moment and call beta a dynamic
correlation in the sense that it goes over 1 plus or minus....

5 year beta's are for practical use at 0 for the bond funds
I've looked at. So 50-50 stocks and bonds ends up with a
beta slightly over .50

Trsy's usually have price advantage in a downturn but overall
interest rises may trump that.
age in bonds, buy-and-hold, 10 year business cycle
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 10:55 pm

Re: Correlations among different types of bonds

Post by lack_ey »

happysteward wrote: Sun Feb 11, 2018 8:33 pm
Treasuries are significantly better than corporates at diversifying the risk of a stock-heavy portfolio.
Doesn’t this argue for a pure treasury bond fund over and above a core bond fund at least as far as portfolio “ballast” goes?
Yes, if that's what you're looking for.

Though really, if you want pure Treasuries, you could also just hold those directly and not use a fund. Save a bit on the expense ratio and target the part of the yield curve you want.
User avatar
Topic Author
grabiner
Advisory Board
Posts: 35307
Joined: Tue Feb 20, 2007 10:58 pm
Location: Columbia, MD

Re: Correlations among different types of bonds

Post by grabiner »

lack_ey wrote: Sun Feb 11, 2018 8:11 pm
grabiner wrote: Sun Feb 11, 2018 7:54 pm (A fund of half stock and half bonds would have 0.71 correlation with stock and 0.71 correlation with bonds if the stock and bond returns were independent.)
My thinking cap is not on right now and I may have missed something, but that seems wrong unless maybe you're also assuming equal volatility of stocks and bonds.
This is correct. A fund with equal volatility from stocks and bonds would have 0.71 correlation with each. With stocks more volatile than bonds, a fund which is half stock and half bonds would have a stronger correlation with stock returns than with bond returns.
Wiki David Grabiner
Theoretical
Posts: 1546
Joined: Tue Aug 19, 2014 10:09 pm

Re: Correlations among different types of bonds

Post by Theoretical »

lack_ey wrote: Sun Feb 11, 2018 9:20 pm
happysteward wrote: Sun Feb 11, 2018 8:33 pm
Treasuries are significantly better than corporates at diversifying the risk of a stock-heavy portfolio.
Doesn’t this argue for a pure treasury bond fund over and above a core bond fund at least as far as portfolio “ballast” goes?
Yes, if that's what you're looking for.

Though really, if you want pure Treasuries, you could also just hold those directly and not use a fund. Save a bit on the expense ratio and target the part of the yield curve you want.
That or play the Brokered/Bank CD game, which can be very rewarding.
Northern Flicker
Posts: 15371
Joined: Fri Apr 10, 2015 12:29 am

Re: Correlations among different types of bonds

Post by Northern Flicker »

livesoft wrote: Sun Feb 11, 2018 8:15 pm The article doesn't seem to mention short-term corporate bonds, yet you mentioned their correlations. Did you calculate them with portfoliovisualizer.com?
It is interesting that short-term corporate bonds have a significantly higher correlation with stocks than intermediate corporates. Shortening corporate bond duration in a stock and bond portfolio may in fact increase portfolio risk. The effect is presumably caused by having less term premium to counterbalance credit risk.
Theoretical
Posts: 1546
Joined: Tue Aug 19, 2014 10:09 pm

Re: Correlations among different types of bonds

Post by Theoretical »

Here's one that goes out to 1991 via the Portfolio Visualizer: http://bit.ly/2CiebYk

Adding TIPS via the PIMCO Real Return Fund (Class I) - 3 years older than the Vanguard Fund - gets you to Feb. 1997: http://bit.ly/2BT6NqR
lazyday
Posts: 3849
Joined: Wed Mar 14, 2007 10:27 pm

Re: Correlations among different types of bonds

Post by lazyday »

livesoft wrote: Sun Feb 11, 2018 8:15 pmThe article doesn't seem to mention short-term corporate bonds, yet you mentioned their correlations.
It's in the table above the text "source: Morningstar Direct"
jalbert wrote: Mon Feb 12, 2018 10:59 pmIt is interesting that short-term corporate bonds have a significantly higher correlation with stocks than intermediate corporates.
The Agg includes Treasuries, not just intermediate corporates.
garlandwhizzer
Posts: 3567
Joined: Fri Aug 06, 2010 3:42 pm

Re: Correlations among different types of bonds

Post by garlandwhizzer »

jalbert wrote;
It is interesting that short-term corporate bonds have a significantly higher correlation with stocks than intermediate corporates. Shortening corporate bond duration in a stock and bond portfolio may in fact increase portfolio risk. The effect is presumably caused by having less term premium to counterbalance credit risk.
This is in contradiction to the action of Vanguard corporate bond funds during the Great Recession of 2007-9 when equities tanked by greater than 50%. The Intermediate Term Corporate Bond Index Fund lost more and was more volatile than the Short Term Corporate Bond Index Fund in the Great Recession. Likewise the Intermediate Term Corporate Investment Grade Fund lost more and was more volatile than the Short Term Corporate Investment Grade Fund. In that severe bear market the correlation between corporate bonds and stocks was increased by increasing corporate bond duration, the opposite of what is stated above.

In Treasuries the opposite was true during that bear market. The Intermediate Term Treasury Fund with significantly positive gains outperformed the more stable Short Term Treasury Fund. Increased duration was rewarded in Treasuries as stocks collapsed. Over long time periods short term instruments of either government or corporate are typically less volatile and lower yielding than intermediate term instruments. In severe bear markets, Intermediate Term Treasuries remain more volatile than Short Term Treasuries but the volatility is to the upside, not the downside which produces higher returns.

The price action of Vanguard funds argues that shortening duration in a corporate bond portfolio does not increase portfolio risk in a bear market, rather the opposite.

Garland Whizzer
User avatar
patrick013
Posts: 3301
Joined: Mon Jul 13, 2015 7:49 pm

Re: Correlations among different types of bonds

Post by patrick013 »

Image

Negative correlation increased in 2008. High Yield was no doubt
strongly correlated. Investment grade corporates should recover
in 2008 but problem there is potential defaults in the financial sector
due to ratings that changed overnight. Wasn't watching investment
grade corporates at the time.

FYI
age in bonds, buy-and-hold, 10 year business cycle
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 10:55 pm

Re: Correlations among different types of bonds

Post by lack_ey »

garlandwhizzer wrote: Tue Feb 13, 2018 12:43 pm
jalbert wrote;
It is interesting that short-term corporate bonds have a significantly higher correlation with stocks than intermediate corporates. Shortening corporate bond duration in a stock and bond portfolio may in fact increase portfolio risk. The effect is presumably caused by having less term premium to counterbalance credit risk.
This is in contradiction to the action of Vanguard corporate bond funds during the Great Recession of 2007-9 when equities tanked by greater than 50%. The Intermediate Term Corporate Bond Index Fund lost more and was more volatile than the Short Term Corporate Bond Index Fund in the Great Recession. Likewise the Intermediate Term Corporate Investment Grade Fund lost more and was more volatile than the Short Term Corporate Investment Grade Fund. In that severe bear market the correlation between corporate bonds and stocks was increased by increasing corporate bond duration, the opposite of what is stated above.

In Treasuries the opposite was true during that bear market. The Intermediate Term Treasury Fund with significantly positive gains outperformed the more stable Short Term Treasury Fund. Increased duration was rewarded in Treasuries as stocks collapsed. Over long time periods short term instruments of either government or corporate are typically less volatile and lower yielding than intermediate term instruments. In severe bear markets, Intermediate Term Treasuries remain more volatile than Short Term Treasuries but the volatility is to the upside, not the downside which produces higher returns.

The price action of Vanguard funds argues that shortening duration in a corporate bond portfolio does not increase portfolio risk in a bear market, rather the opposite.

Garland Whizzer
Sure, it should be clear if you think about what happens when the credit spread changes. A 1% increase in spread will hurt a corporate bond with 10 years until maturity/call a lot more than a corporate bond with 1 year maturity remaining.

Furthermore, in a downside event for credit, longer-term corporate bonds are then less likely going to be called (no money to pay off early, no incentive to refinance at higher rates), so the duration can increase as the spreads spike up.

I definitely wouldn't count on extending duration in corporate bonds to reduce correlation with equities. Now, when credit spreads are relatively stable, term risk is driving the difference in returns, and term risk is being negatively correlated with equity market risk, you can get that effect, but that's a few too many conditions there.
Northern Flicker
Posts: 15371
Joined: Fri Apr 10, 2015 12:29 am

Re: Correlations among different types of bonds

Post by Northern Flicker »

This is in contradiction to the action of Vanguard corporate bond funds during the Great Recession of 2007-9 when equities tanked by greater than 50%...
I would agree with this and lack_ey's subsequent point. I think the presented sample correlations just exhibit the kind of sample bias that is pervasive in asset return data.
Post Reply