Total Bond vs IT Treasuries vs GNMAs in a 3-fund portfolio

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Total Bond vs IT Treasuries vs GNMAs in a 3-fund portfolio

Post by Northern Flicker »

McQ wrote: Mon May 16, 2022 4:15 pm
Northern Flicker wrote: Mon May 16, 2022 1:03 am I still consider the lower correlation of treasuries with equities relative to the correlation of corporate bonds with equities to be a key issue.
...
To your point about correlations, Northern Flicker, here is a paper by Kizer et al. that backs up your concerns: https://papers.ssrn.com/sol3/papers.cfm ... id=3147005
Thanks for the reference. The article seems to draw the conclusion that credit neither improves nor degrades portfolio efficiency when added to a portfolio of equities and treasuries. This suggests that the main issue is establishing the asset allocation to be reflective of one’s desired risk level.

That has been my view in the past, but various backtests I’ve looked at more recently suggests to me that the portfolio exclusively with treasuries is very slightly more efficient. This is from the observation that over different periods a treasury-only bond portfolio packaged with equities may slightly overperform or slightly underperform the same portfolio with a more diversified bond index that includes credit. But in all scenarios, the treasury-only portfolio seems to provide superior downside protection.

Generally, I don’t think it is an error to use a total bond index portfolio, but don’t see any reason why any boglehead proponents should be troubled by the idea of using an intermediate treasury index fund like FUAMX or VGIT as the bond index fund in a 3-fund portfolio recommendation. I consider the available evidence to suggest that the treasury-only bond index fund leads to a portfolio that is slightly superior to using a total bond index fund.

I would hesitate to use nominal bond funds exclusively for portfolios with low equity allocations, but that is a separate discussion.
000
Posts: 8211
Joined: Thu Jul 23, 2020 12:04 am

Re: Total Bond vs IT Treasuries vs GNMAs in a 3-fund portfolio

Post by 000 »

McQ wrote: Mon May 16, 2022 4:23 pm Yes and no, 000. The US bond market was a very different place before 1918, but not in precisely the way you suggested. It wasn’t that return-maximizing investors bought bonds instead of stocks; bonds were bought then, as now, for a stable income at minimum risk.

The difference was the near-complete absence of Treasury securities. That market had gone moribund after the Civil War refinancing was completed circa 1879. Not much trading, not much liquidity, not much face value available to trade, and unattractive rates (Treasuries were locked up in bank vaults to gain circulation privileges). It’s not really possible to compute a meaningful yield spread in this era.

There were municipal bonds then, but these had the same liquidity problems, etc. Railroad bonds were available in the tens of millions, were blue chip securities, paid a market rate of interest, and were essentially, the only fixed income game in town until Liberty bonds debuted in 1917, after which the modern Treasury market gradually began to take shape (an evolution not completed until the 1970s, BTW).

PS re defaults: except for the 1890s, and later the 1930s, defaults on blue chip railroad bonds were rare, and there was generally a work out where new securities equal to the par value eventually got paid out to the bondholders, albeit with a gap in income and some loss of security. "Blue chip" is the constraint here: if these were carpet bagger Southern railroad bonds created out of thin air through legislative bribery after the Civil War, these mostly went bust after 1873. And if it was the Erie railroad, you could be in for a rough ride, but no one would call "the Scarlet Woman of Wall Street" a blue-chip security ...
Thank you. I think the most fascinating thing about all of this is how few non-overlapping historical periods that are somewhat similar to our current financial environment we actually have.
Topic Author
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Total Bond vs IT Treasuries vs GNMAs in a 3-fund portfolio

Post by Northern Flicker »

000 wrote: I think the most fascinating thing about all of this is how few non-overlapping historical periods that are somewhat similar to our current financial environment we actually have.
That may still be true in another couple of hundred years. Apparently, the first known example of a bond was a piece of stone onto which terms were engraved in ancient Mesopotamia about 4400 years ago:

http://bondfunds.com/education/a-brief- ... investing/

So don’t ever say that the contractual terms of a bond cannot be cast in stone!
McQ
Posts: 1425
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Total Bond vs IT Treasuries vs GNMAs in a 3-fund portfolio

Post by McQ »

Hello Northern Flicker: I don’t want to challenge your endorsement of intermediate Treasuries per se. But I do want to deliver a caution about abbreviated backtesting.

Here is the caution: do your results show the superiority of Intermediate Treasuries over Total Bond: or, do they show the superiority of the duration characteristic of IT, in that era, over the duration characteristic of Total Bond, in that era?

As I hope you will agree, although both are characterized as “intermediate,” the duration of Total Bond (BND) is highly unlikely to be identical to any Intermediate Treasury fund (VGIT) at any juncture. And the deviation may not be found just to the right of the decimal place but in the integer values as well.

When bond prices are rising, a bond fund with a slight duration excess will outperform its foil. When bond prices are falling, the fund with the lower duration will outperform.

Here are two charts from a different paper, where I had occasion to extend the total bond index back to 1900. https://papers.ssrn.com/sol3/papers.cfm ... id=3947293

The charts reset the bond series at the boundary of bull / bear moves. This first chart separates the bond bull market of 1921 to 1946 from what came before and after. Long intermediate is bonds with a maturity between 6 and 10 years; short intermediate is a maturity of about 5 years (generally, the SBBI series).

Image

When bond prices are rising, a total bond index, with its significant exposure to long bonds, outperforms a short intermediate series (shaded area). When bond prices are falling, a short intermediate bond series will outperform a total bond series, and a longer intermediate series will match it.

Next chart is similar, but covers only the Barcap era. Same finding, using only a single intermediate series: total bond underperforms intermediate in the bear cycle, outperforms in the bull cycle. Duration rules.

Image

The question I leave you with: do your results show the superiority of Treasuries per se, or the superiority of a bond fund having the exact duration of intermediate Treasuries at that time, relative to the somewhat different duration of Total Bond at that time?
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Topic Author
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Total Bond vs IT Treasuries vs GNMAs in a 3-fund portfolio

Post by Northern Flicker »

All backtests are biased. I think you made a good case that a long backtest starting say in 1917 would have been less representative.

Comparing the duration of a treasury portfolio to the total bond index is challenging because treasuries have a deterministic duration and MBS and corporate bonds have (stochastic) expected durations that vary over time.

Generally, though, the total bond index has had a duration that was a bit longer than the treasury fund VFITX/VFIUX. This actually favored the total bond index in the periods where the portfolio with treasuries had a slightly higher return. This suggests that variations in credit spreads or MBS effects were the more distinguishing changes for total bond during the period.

But I think the results are reasonably compelling that the 60/40 portfolio with treasuries has somewhat less downside risk. I would turn the question around and ask for evidence to expect that the additional risk from the inclusion of other bonds in the bond index would be rewarded when combined in a portfolio with equities. That is the question that needs to be answered.

Of course we always have to assume that we are just observing an outcome and not a generalizable effect with a measured portfolio return in a backtest.
xxd091
Posts: 494
Joined: Sun Aug 21, 2011 4:41 am
Location: UK

Re: Total Bond vs IT Treasuries vs GNMAs in a 3-fund portfolio

Post by xxd091 »

Would agree re the results of treasuries over bond funds
I found as I got older that running a bond ladder became tiresome and a bond fund did almost the same job with no input needed from me except overview
Now bond fund is part of a 3 fund portfolio that needs very little hands on-suits the older retired investor
xxd091
User avatar
Doc
Posts: 10607
Joined: Sat Feb 24, 2007 12:10 pm
Location: Two left turns from Larry

Re: Total Bond vs IT Treasuries vs GNMAs in a 3-fund portfolio

Post by Doc »

Northern Flicker wrote: Tue May 17, 2022 12:34 am But I think the results are reasonably compelling that the 60/40 portfolio with treasuries has somewhat less downside risk. I would turn the question around and ask for evidence to expect that the additional risk from the inclusion of other bonds in the bond index would be rewarded when combined in a portfolio with equities. That is the question that needs to be answered.
Finally someone has at least raised the overall performance question of different bond types in portfolios which also contain equities.

Thanks.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.
Topic Author
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Total Bond vs IT Treasuries vs GNMAs in a 3-fund portfolio

Post by Northern Flicker »

I was thinking further about the point raised by McQ regarding duration. There have been many discussions on BH concerning the behavior of different bond subclasses when packaged with equities, including throughout this thread. A point that occurred to me that I’ve not seen mentioned before is that treasuries not only diversify downside equity risk more effectively than other nominal bond classes, but they accomplish it with a lower duration.

Generally, treasuries with TERM exposure are more effective at protecting against downside risk of equities during downturns. This may not be true when inflation is accelerating, a topic on many minds today, but that discussion is not very relevant to comparing treasuries with total bond for equity risk diversification. The backtests in the original posting provide what I find to be compelling evidence that treasuries with a slightly lower duration than total bond (5.2 vs around 6.5) still provide better downside protection in a 60/40 portfolio for instance.

This means that the investor can achieve the downside protection with less term risk. Of course total bond has mortgages and credit which have drivers of yield changed that are not fully correlated with term risk. Credit risk may in fact be negatively correlated with term risk (due to positive equity correlation). So when rates are rising, total bond needs credit spreads to narrow enough to counterbalance the longer duration and greater exposure to term risk. This may or may not happen.

So if rates are falling due to a slowing economy, treasuries provide superior downside protection even with a lower duration. When rates are rising during a growing economy (or during accelerating inflation) the lower duration will be preferred, and the investor needs credit spreads to cooperate by enough for total bond to avoid a heads-I-win-tails-you-lose scenario of underperforming the shorter duration treasuries on both the downturn and its recovery.

I think it should be attractive to investors to provide equivalent or superior protection against downside equity risk with a shorter duration bond portfolio.
Post Reply