Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
longinvest
Posts: 3197
Joined: Sat Aug 11, 2012 8:44 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by longinvest » Fri Jan 11, 2019 5:53 pm

I guess that Vanguard's active bond fund managers are simply boosting income by buying premium bonds, while keeping true to their target benchmark on a total return basis.

I wouldn't be surprised to learn that active bond fund investors have a preference for higher income. This would match the behavior of income investors.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic / international) stocks / (nominal / inflation-indexed) bonds | VCN/VXC/VLB/ZRR

dcabler
Posts: 747
Joined: Wed Feb 19, 2014 11:30 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by dcabler » Sat Jan 12, 2019 7:39 am

siamond wrote:
Fri Jan 11, 2019 5:36 pm
I ran a few more tests and swapped a couple of messages with longinvest.

The bottomline is that Vanguard treasury bonds which are NOT index funds (e.g. VUSTX, VFITX, VFISX) seem to be more active than I would have expected, and their price series have a life of their own... It seems that they end up with similar total returns using rather different means than indices (or corresponding index funds). Trouble is we have less than a decade of returns for true bond index funds (e.g. VLGSX, VSIGX, VSBSX), so it isn't terribly significant to check the simulator's numbers against such limited reality. And corresponding indices (for IT/ST treasuries) do not have a PR (price-only) data series available in Morningstar.

I don't know what else to do to run a solid sanity check about the new income/capital breakdown from the simulator. I'd be inclined to just trust longinvest and call it a day...
It's not much help, but Fido's series give you a couple of extra years as these go back to early Jan 2008.
FUAMX (ITT) don't use the search feature in Morningstar. This is one of the funds that changed names recently and M* hasn't completely updated things. However, if pull up another fund and just change the URL to FUAMX, then you get the info)
FNBGX (LTT)
FXNAX (STT)

User avatar
siamond
Posts: 4410
Joined: Mon May 28, 2012 5:50 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by siamond » Sat Jan 12, 2019 1:02 pm

dcabler wrote:
Sat Jan 12, 2019 7:39 am
It's not much help, but Fido's series give you a couple of extra years as these go back to early Jan 2008.
Interesting idea! I looked at treasury index funds from Fidelity (e.g. FUMBX, FUAMX, FNBGX), they were created in 2005, so yes, we get more history. I started to check FUAMX (the IT fund), the total-returns match the corresponding index quite well, but... the price series started by the end of 2017, aarg. Which we can more directly observe by checking the price chart on Morningstar. Same for the other funds.

This loss of history is probably a side effect of the ongoing merging of asset classes. I am stuck again, hmpf.

User avatar
siamond
Posts: 4410
Joined: Mon May 28, 2012 5:50 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by siamond » Sat Jan 12, 2019 6:09 pm

After some detective work (using 'time machine' copies of Morningstar Web pages!), I was able to reconstruct the data series for the Fidelity index funds. I used:
* FLBAX - Fidelity Spartan Long-Term Treasury Bond Index Fund
* FIBAX - Fidelity Spartan Intermediate-Term Treasury Bond Index Fund

Both funds share class was 'Advantage' (ER = 0.07), they were created by the end of 2005 and closed early Nov-18. Which means that I couldn't reconstruct the 2018 returns for those funds, so I used instead the newer 'class-less' Fidelity funds for 2018 (FNBGX and FUAMX).

I compared to the corresponding Vanguard index funds (inception end of 2009), which have the same ER (0.07):
* VLGSX - Vanguard Long-Term Treasury Index Fund (Admiral)
* VSIGX - Vanguard Intermediate-Term Treasury Index Fund (Admiral)

To fill the blanks and ease the comparison, I equated the (capital and total) returns of the Vanguard funds to Fidelity for 2006-2009. Then I compared to the index data series from Barclays ("/X" notation, PR/TR for LTT, TR only for ITT -PR missing-). And compared to the capital and total returns from longinvest's latest model ("/M" notation). And assembled the followed growth charts. I hope I didn't mess up with the inputs... Click on the image to see larger charts.

There is a lot going on in those charts... My bottomline is that it seems more satisfying for total-returns than for prices, but when we see the price variations between Fidelity and Vanguard index funds, the price model might actually be fairly ok. That is, for those few years where we can compare to actuals.

EDIT: shoot, the Intermediate-Term comparison isn't valid... Fidelity FIBAX/FUAMX follows Barclays U.S. 5-10 Year Treasuries (the ITT index I used in the charts below) while Vanguard VSIGX follows Barclays U.S. 3-10 Year Treasuries. Missed that... :(

Image

Image

dcabler
Posts: 747
Joined: Wed Feb 19, 2014 11:30 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by dcabler » Sun Jan 13, 2019 6:40 am

siamond wrote:
Sat Jan 12, 2019 6:09 pm
After some detective work (using 'time machine' copies of Morningstar Web pages!), I was able to reconstruct the data series for the Fidelity index funds. I used:
* FLBAX - Fidelity Spartan Long-Term Treasury Bond Index Fund
* FIBAX - Fidelity Spartan Intermediate-Term Treasury Bond Index Fund

Both funds share class was 'Advantage' (ER = 0.07), they were created by the end of 2005 and closed early Nov-18. Which means that I couldn't reconstruct the 2018 returns for those funds, so I used instead the newer 'class-less' Fidelity funds for 2018 (FNBGX and FUAMX).

I compared to the corresponding Vanguard index funds (inception end of 2009), which have the same ER (0.07):
* VLGSX - Vanguard Long-Term Treasury Index Fund (Admiral)
* VSIGX - Vanguard Intermediate-Term Treasury Index Fund (Admiral)

To fill the blanks and ease the comparison, I equated the (capital and total) returns of the Vanguard funds to Fidelity for 2006-2009. Then I compared to the index data series from Barclays ("/X" notation, PR/TR for LTT, TR only for ITT -PR missing-). And compared to the capital and total returns from longinvest's latest model ("/M" notation). And assembled the followed growth charts. I hope I didn't mess up with the inputs... Click on the image to see larger charts.

There is a lot going on in those charts... My bottomline is that it seems more satisfying for total-returns than for prices, but when we see the price variations between Fidelity and Vanguard index funds, the price model might actually be fairly ok. That is, for those few years where we can compare to actuals.

EDIT: shoot, the Intermediate-Term comparison isn't valid... Fidelity FIBAX/FUAMX follows Barclays U.S. 5-10 Year Treasuries (the ITT index I used in the charts below) while Vanguard VSIGX follows Barclays U.S. 3-10 Year Treasuries. Missed that... :(

Image


Image

Isn't the total return data from longinvest's spreadsheet originally calibrated (or at least compared) to the somewhat-active versions of Vanguard's bond funds, not their newer passive counterparts? Not sure how much difference trying to change that would make when trying to compare to either the Vanguard or Fido series, but maybe something to consider if you're just trying to check the methodology.

Topic Author
longinvest
Posts: 3197
Joined: Sat Aug 11, 2012 8:44 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by longinvest » Sun Jan 13, 2019 9:51 am

dcabler wrote:
Sun Jan 13, 2019 6:40 am
Isn't the total return data from longinvest's spreadsheet originally calibrated (or at least compared) to the somewhat-active versions of Vanguard's bond funds, not their newer passive counterparts?
The total return data in the simulator spreadsheet is directly derived by simulating ladder-like* bond funds using available historical yields (Shiller, NBER, and FRED), as described in the first post of this thread, without any calibration whatsoever.

* With the exception of the "EqPar Zero fund (30 to 21-year)" which is equal-par weighted.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic / international) stocks / (nominal / inflation-indexed) bonds | VCN/VXC/VLB/ZRR

User avatar
siamond
Posts: 4410
Joined: Mon May 28, 2012 5:50 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by siamond » Sun Jan 13, 2019 10:45 am

dcabler wrote:
Sun Jan 13, 2019 6:40 am
Isn't the total return data from longinvest's spreadsheet originally calibrated (or at least compared) to the somewhat-active versions of Vanguard's bond funds, not their newer passive counterparts? Not sure how much difference trying to change that would make when trying to compare to either the Vanguard or Fido series, but maybe something to consider if you're just trying to check the methodology.
There is a calibration of sorts to the indices followed by the Vanguard funds, e.g. the funds being currently modeled match the range of (treasury) maturities used by the index of relevance. For long-term bonds, both Vanguard and Fidelity follow the same index and the bond fund 30-11 model is comparable. Now, as we discovered, the active flavor of the Vanguard seems to do a fairly good job of tracking total returns of its index, while its price trajectory displayed a life of their own... While the passive flavor of both Vanguard and Fidelity got closer to the price trajectory of the index.

My silly mistake yesterday was to assume that the same was true for intermediate-term bonds, but as I discovered afterwards, Vanguard and Fidelity made different choices of IT index. Which means that the 10-4 bond fund model (calibrated for the index Vanguard follows) wasn't relevant for the comparison with Fidelity. Making my IT comparison charts basically meaningless. Oh well, I tried. :|

dcabler
Posts: 747
Joined: Wed Feb 19, 2014 11:30 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by dcabler » Sun Jan 13, 2019 10:45 am

longinvest wrote:
Sun Jan 13, 2019 9:51 am
dcabler wrote:
Sun Jan 13, 2019 6:40 am
Isn't the total return data from longinvest's spreadsheet originally calibrated (or at least compared) to the somewhat-active versions of Vanguard's bond funds, not their newer passive counterparts?
The total return data in the simulator spreadsheet is directly derived by simulating ladder-like* bond funds using available historical yields (Shiller, NBER, and FRED), as described in the first post of this thread, without any calibration whatsoever.

* With the exception of the "EqPar Zero fund (30 to 21-year)" which is equal-par weighted.
OK, but they were compared as I see when I go back through the this thread. Maybe the question is really for siamond. At some point earlier in the thread, siamond picked up on longinvest's work to use the returns generated from your spreadsheet as a proxy for the various vanguard funds in the simba spreadsheet before their actual inception date. At some point in the 2016 timeframe, it was determined that the 10-4 model was a decent substitute for the index that VFITX more or less follows...

So if the purpose is to just compare the methodology for generating price returns vs. reality, and if the Fido data goes back the farthrest, is the 10-4 model close enough, or should there be a modification to the spreadsheet to add something that better approximates the Barclays index that Fido uses?
Second, but related question, is whether Simba should switch over to the pure index funds instead of the managed ones (sorry - off topic)

dcabler
Posts: 747
Joined: Wed Feb 19, 2014 11:30 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by dcabler » Sun Jan 13, 2019 10:49 am

siamond wrote:
Sun Jan 13, 2019 10:45 am
dcabler wrote:
Sun Jan 13, 2019 6:40 am
Isn't the total return data from longinvest's spreadsheet originally calibrated (or at least compared) to the somewhat-active versions of Vanguard's bond funds, not their newer passive counterparts? Not sure how much difference trying to change that would make when trying to compare to either the Vanguard or Fido series, but maybe something to consider if you're just trying to check the methodology.
There is a calibration of sorts to the indices followed by the Vanguard funds, e.g. the funds being currently modeled match the range of (treasury) maturities used by the index of relevance. For long-term bonds, both Vanguard and Fidelity follow the same index and the bond fund 30-11 model is comparable. Now, as we discovered, the active flavor of the Vanguard seems to do a fairly good job of tracking total returns of its index, while its price trajectory displayed a life of their own... While the passive flavor of both Vanguard and Fidelity got closer to the price trajectory of the index.

My silly mistake yesterday was to assume that the same was true for intermediate-term bonds, but as I discovered afterwards, Vanguard and Fidelity made different choices of IT index. Which means that the 10-4 bond fund model (calibrated for the index Vanguard follows) wasn't relevant for the comparison with Fidelity. Making my IT comparison charts basically meaningless. Oh well, I tried. :|
Yep, that's what I recalled and what I saw when I went back to the earlier postings in this thread.

User avatar
siamond
Posts: 4410
Joined: Mon May 28, 2012 5:50 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by siamond » Sun Jan 13, 2019 11:00 am

dcabler wrote:
Sun Jan 13, 2019 10:45 am
So if the purpose is to just compare the methodology for generating price returns vs. reality, and if the Fido data goes back the farthrest, is the 10-4 model close enough, or should there be a modification to the spreadsheet to add something that better approximates the Barclays index that Fido uses?
Well, yes, I could ask longinvest to add a 10-6 model which would match better the index followed by the Fidelity IT fund. But given that Fidelity only adds a few years of real-life record, it is really not worth the effort.
dcabler wrote:
Sun Jan 13, 2019 10:45 am
Second, but related question, is whether Simba should switch over to the pure index funds instead of the managed ones (sorry - off topic)
Yes, this is a legitimate question. I am not too hot about making a spliced series with years of active fund (e.g. VUSTX) followed by more recent years of passive fund (e.g. VLGSX). What we could do is to switch to a series splicing the bond model of relevance, then the index until 2009, then the corresponding passive fund. Such series would undoubtedly be more consistent. We would lose more than 20 years of real-life history though. Plus, I wonder how many Bogleheads actually use those truly passive bond funds as opposed to the more well-known active funds. Feedback welcome, it is clearly a valid question, but I am not too sure such a change would be worth it.

Topic Author
longinvest
Posts: 3197
Joined: Sat Aug 11, 2012 8:44 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by longinvest » Sun Jan 13, 2019 11:08 am

siamond wrote:
Sun Jan 13, 2019 11:00 am
Well, yes, I could ask longinvest to add a 10-6 model which would match better the index followed by the Fidelity IT fund. But given that Fidelity only adds a few years of real-life record, it is really not worth the effort.
It would be quite easy, actually, to add a 10-6 model. Want me to do it?

Self-correction would extend as far as the 10-4 model, as FRED provides both 3 and 5 year constant maturity yields since 1954.
Last edited by longinvest on Sun Jan 13, 2019 11:22 am, edited 2 times in total.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic / international) stocks / (nominal / inflation-indexed) bonds | VCN/VXC/VLB/ZRR

dcabler
Posts: 747
Joined: Wed Feb 19, 2014 11:30 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by dcabler » Sun Jan 13, 2019 11:13 am

siamond wrote:
Sun Jan 13, 2019 11:00 am
dcabler wrote:
Sun Jan 13, 2019 10:45 am
So if the purpose is to just compare the methodology for generating price returns vs. reality, and if the Fido data goes back the farthrest, is the 10-4 model close enough, or should there be a modification to the spreadsheet to add something that better approximates the Barclays index that Fido uses?
Well, yes, I could ask longinvest to add a 10-6 model which would match better the index followed by the Fidelity IT fund. But given that Fidelity only adds a few years of real-life record, it is really not worth the effort.
dcabler wrote:
Sun Jan 13, 2019 10:45 am
Second, but related question, is whether Simba should switch over to the pure index funds instead of the managed ones (sorry - off topic)
Yes, this is a legitimate question. I am not too hot about making a spliced series with years of active fund (e.g. VUSTX) followed by more recent years of passive fund (e.g. VLGSX). What we could do is to switch to a series splicing the bond model of relevance, then the index until 2009, then the corresponding passive fund. Such series would undoubtedly be more consistent. We would lose more than 20 years of real-life history though. Plus, I wonder how many Bogleheads actually use those truly passive bond funds as opposed to the more well-known active funds. Feedback welcome, it is clearly a valid question, but I am not too sure such a change would be worth it.
Thanks siamond - I'll move the discussion on the bond funds over to the Simba thread and put the question there.

User avatar
siamond
Posts: 4410
Joined: Mon May 28, 2012 5:50 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by siamond » Sun Jan 13, 2019 11:21 am

longinvest wrote:
Sun Jan 13, 2019 11:08 am
siamond wrote:
Sun Jan 13, 2019 11:00 am
Well, yes, I could ask longinvest to add a 10-6 model which would match better the index followed by the Fidelity IT fund. But given that Fidelity only adds a few years of real-life record, it is really not worth the effort.
It would be quite easy, actually, to add a 10-6 model. Want me to do it?
For the sake of trying to further validating the capital return (price) model, I really don't think it's worth adding it (note that M* doesn't provide PR series for IT indices, neither of them). Now, for the sake of having a reference to extend the history of the Fidelity IT fund, this might present *some* value? Not that I have any plan to add such a data series to Simba, but maybe one day, we'll find a good reason to use it, e.g. to help compare 10-5 to 10-3 index strategies and help some investors to choose between Vanguard and Fidelity funds. Up to you, but my motivation is rather low, quite frankly!

Topic Author
longinvest
Posts: 3197
Joined: Sat Aug 11, 2012 8:44 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by longinvest » Sun Jan 13, 2019 12:03 pm

I have uploaded version 1.21 of the Bond Fund Simulator spreadsheet.

It is available online and to download from the links in the first post.

Main changes:
  • Added Bond Fund (10 to 6-year).
As usual, comments are welcome.

Enjoy!
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic / international) stocks / (nominal / inflation-indexed) bonds | VCN/VXC/VLB/ZRR

Topic Author
longinvest
Posts: 3197
Joined: Sat Aug 11, 2012 8:44 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by longinvest » Sun Jan 13, 2019 11:58 pm

I have uploaded version 1.22 of the Bond Fund Simulator spreadsheet.

It is available online and to download from the links in the first post.

Main changes:
  • Added Bond Fund (5 to 2-year).
As usual, comments are welcome.

Enjoy!
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic / international) stocks / (nominal / inflation-indexed) bonds | VCN/VXC/VLB/ZRR

User avatar
siamond
Posts: 4410
Joined: Mon May 28, 2012 5:50 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by siamond » Tue Jan 15, 2019 11:14 am

Longinvest has been steadily adding new bond fund models to the simulator. It seems to me that it would be a good time to develop some generic set of rules as well as specific recommendations when trying to best match a given real-life bond fund to a given model. Case in point, the main bond funds from Vanguard and Fidelity. Let me start by a few which are top of mind for the Simba backtesting spreadsheet.

Vanguard Total Bond Market Index Fund (VBTLX/VBMFX):
- passive index fund, mix of treasuries, government and corporates
- benchmark: Bloomberg Barclays U.S. Aggregate Bond Index
- current average maturity: 8.4 years
- current maturity distribution is wide, including a good deal of (~40%) of bonds maturing in 5 years or less

Vanguard Short-Term Treasury Index Fund (VSBSX):
- passive index fund, treasuries only
- benchmark: Bloomberg Barclays U.S. Treasury 1-3 Year Index
- current average maturity: 2.0 years
- current maturity distribution is entirely made of of bonds maturing in 3 years or less
(the corresponding Fidelity index fund, FUMBX, is anchored on a 1-5 yrs index, confusing!)

Vanguard Intermediate-Term Treasury Index Fund (VSIGX):
- passive index fund, treasuries only
- benchmark: Bloomberg Barclays U.S. Treasury 3-10 Year Index
- current average maturity: 5.6 years
- current maturity distribution is ~50% 3-5 years and ~50% 5-10 years
(the corresponding Fidelity index fund, FUAMX, is anchored on a 5-10 yrs index, confusing!)

Vanguard Long-Term Treasury Index Fund (VLGSX):
- passive index fund, treasuries only
- benchmark: Bloomberg Barclays U.S. Long Treasury Index
- current average maturity: 25.1 years
- current maturity distribution is very skewed towards very long term (93% 20-30 yrs; 7% 10-20 yrs)
(the corresponding Fidelity index fund, FNBGX, is mercifully anchored on the same index)

Vanguard Short-Term Treasury Fund (VFISX/VFIRX):
- active fund, treasuries only
- benchmark: Bloomberg Barclays U.S. 1-5 Year Treasury Bond Index
- current average maturity: 2.2 years
- current maturity distribution is 90% made of of bonds maturing in 3 years or less
(NOT the same benchmark as VSBSX although current characteristics are quite similar, confusing!)

Vanguard Intermediate-Term Treasury Fund (VFITX/VFIUX):
- active fund, treasuries only
- benchmark: Bloomberg Barclays U.S. 5-10 Year Treasury Bond Index
- current average maturity: 6.0 years
- current maturity distribution is ~40% 3-5 years and ~60% 5-10 years
(NOT the same benchmark as VSIGX although current characteristics are somewhat similar, confusing!)

Vanguard Long-Term Treasury Fund (VUXTX/VUSUX):
- active fund, treasuries only
- benchmark: Bloomberg Barclays U.S. Long Treasury Bond
- current average maturity: 24.8 years
- current maturity distribution is very skewed towards very long term (91% 20-30 yrs; also includes a few % of short-term maturities)
(mercifully close to its index companion fund, VLGSX, except for the few short-term bonds)

Feedback welcome from bond experts...

User avatar
Kevin M
Posts: 10398
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by Kevin M » Tue Jan 15, 2019 12:20 pm

It would seem that an index fund is going to be easier to match with a model. By definition, the makeup of an actively-managed bond fund is going to change over time in an unpredictable way, so it would seem that it would be more difficult to model.

A few years ago, I compared the holdings of the VG intermediate-term Treasury fund (not the index fund) at a few different points in time, and the distribution of maturities was quite different at different times.

Nevertheless, I believe it's been shown that the appropriate longinvest model does a decent job of matching the performance (total return) of the corresponding actively-managed Vanguard bond fund. My understanding is that Vanguard actively-managed bond funds are relatively passive compared to some other actively-managed bond funds; i.e., they don't make really big bets that deviate too much from the benchmark.

Sorry I don't have anything more concrete to offer.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

dcabler
Posts: 747
Joined: Wed Feb 19, 2014 11:30 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by dcabler » Tue Jan 15, 2019 3:46 pm

Kevin M wrote:
Tue Jan 15, 2019 12:20 pm
It would seem that an index fund is going to be easier to match with a model. By definition, the makeup of an actively-managed bond fund is going to change over time in an unpredictable way, so it would seem that it would be more difficult to model.

A few years ago, I compared the holdings of the VG intermediate-term Treasury fund (not the index fund) at a few different points in time, and the distribution of maturities was quite different at different times.

Nevertheless, I believe it's been shown that the appropriate longinvest model does a decent job of matching the performance (total return) of the corresponding actively-managed Vanguard bond fund. My understanding is that Vanguard actively-managed bond funds are relatively passive compared to some other actively-managed bond funds; i.e., they don't make really big bets that deviate too much from the benchmark.

Sorry I don't have anything more concrete to offer.

Kevin
I've noticed that, too, even with index bond funds. Any idea why? It would seem that an index bond fund isn't a perfect conveyor belt. I tried to find construction rules some time ago but couldn't.

kolea
Posts: 1320
Joined: Fri Jul 11, 2014 5:30 pm
Location: Maui and Columbia River Gorge

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by kolea » Tue Jan 15, 2019 5:23 pm

dcabler wrote:
Tue Jan 15, 2019 3:46 pm
Kevin M wrote:
Tue Jan 15, 2019 12:20 pm
It would seem that an index fund is going to be easier to match with a model. By definition, the makeup of an actively-managed bond fund is going to change over time in an unpredictable way, so it would seem that it would be more difficult to model.

A few years ago, I compared the holdings of the VG intermediate-term Treasury fund (not the index fund) at a few different points in time, and the distribution of maturities was quite different at different times.

Nevertheless, I believe it's been shown that the appropriate longinvest model does a decent job of matching the performance (total return) of the corresponding actively-managed Vanguard bond fund. My understanding is that Vanguard actively-managed bond funds are relatively passive compared to some other actively-managed bond funds; i.e., they don't make really big bets that deviate too much from the benchmark.

Sorry I don't have anything more concrete to offer.

Kevin
I've noticed that, too, even with index bond funds. Any idea why? It would seem that an index bond fund isn't a perfect conveyor belt. I tried to find construction rules some time ago but couldn't.
The Barclay US Aggregate index is silent on how maturities are distributed. It is more specific on the distribution of quality and source of the securities. Maturity distribution can have a huge effect on how a fund performs. For instance, BND, the Vanguard fund, is a barbell, with a bit of weight out at longer maturities. They do this to juice the returns. But that is a lot of latitude for a bond fund manager and it really blurs the distinction between an active fund and a passive fund.

I have also been quite interested in simulating a bond fund and wrote a simulator in Java. As this thread discusses, it became quite obvious to me that the key to getting the fund to be at all accurate is to have the full historical yield curve. For me, I am less interested in simulating a specific bond fund and more interested in seeing how a bond fund behaves when things change, for instance what happens as the yield curve steepens or flattens? Stuff like that. But I also use my simulator as input to a RIP tool I wrote and it would be nice to at least have the slope of the yield curve between the 1-year and 10-year points, as a function of time. But the full yield curve, going back to 1926 or so, would be great!
Kolea (pron. ko-lay-uh). Golden plover.

Topic Author
longinvest
Posts: 3197
Joined: Sat Aug 11, 2012 8:44 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by longinvest » Wed Jan 16, 2019 6:17 pm

siamond wrote:
Tue Jan 15, 2019 11:14 am
It seems to me that it would be a good time to develop some generic set of rules as well as specific recommendations when trying to best match a given real-life bond fund to a given model.
I suggest to keep things simple:

1- Treasury index funds: match the fund to the model that contains the same maturities.

Example: Vanguard Long-Term Treasury Index Fund Admiral Shares (VLGSX) "invests in debt issued by the U.S. Treasury (not including inflation-protected bonds) and U.S. government agencies, as well as corporate or dollar-denominated foreign debt guaranteed by the U.S. government, with maturities greater than 10 years". As a consequence, I would choose the model that includes all bonds with a maturity equal or higher than 10 years. This would be Bond Fund (30 to 11-year), because the 11-year bond is kept for one year and sold when it reaches 10 years to maturity.

2- Total Bond Market index fund: match the fund to the Bond Fund (10 to 2-year) model. Various reasons:
  • The total bond market contains a wide variety of bonds: corporates, Treasuries, Government mortgage-backed, etc. These bonds cover all maturities above 1 year. Yet, the weighting of short-term bonds is high. As a result, the average duration of the total bond market is 6 years. It's half a year longer than the 5.5 year average duration of Bond Fund (10 to 2 year). But, the wide variety of bonds it contains gives it a smoother behavior than implied by its duration. We've seen this in 2008-2009, when corporates dampened the volatility of Treasuries, making the total market quite smooth. By using a model based on default-less Treasuries with a slightly shorter duration, the volatility is likely to be somewhat similar.
  • To my eye, when looking visually at a logarithmic growth chart of Vanguard's Total Bond Market Index Fund (VBMFX) since inception and of Bond Fund (10 to 2-year), they seem to share a relatively similar volatility. Bund Fund 10-2 is somewhat more volatile (e.g. in 2008-2009), but it's similar enough. They also share close-enough returns. It's not perfect, but I don't see an obviously better match among the models.
  • The total bond market can be considered as a good default bond fund, for an investor, today. Such a fund didn't exist in 1871, but it would be reasonable to assume that a bond investor of the time would have maintained a bond ladder. Maybe a super sophisticated investor would have sold his bond one year prior to maturity, as the Bond Fund (10 to 2-year) model does.
  • Bond Fund (10 to 2-year) is the only "fund" model with self-correcting returns going back to 1871. There's also Bond Ladder (10 to 1-year), but it's a "ladder", not a "fund". (I use "fund" when bonds aren't held until maturity). Other fund models are also simulated back to 1871, but they only gain the self-correcting property in later years.
3- Treasury funds (not indexed): when necessary, because of a lack of an index fund, use the model covering the same maturities as the fund.

4- Other Bond funds: don't match to a model.

One annoying issue that remains is to decide what is:
  • Short-term: is it 1 to 3, or 1 to 5 years?
  • Intermediate-term: is it 3 to 10, or 5 to 10 years?
  • Long-term: I think that most funds define this as 10 years or more.
My personal preference would be 1 to 5 years for short term and, correspondingly, 5 to 10 years for intermediate term. I find 1 to 3 years too narrow (only 2 rungs in the fund model). This choice would result into the following matches:
  • Short-term Treasury index: Bond Fund (5 to 2-year)
  • Intermediate-term Treasury index: Bond Fund (10 to 6-year)
  • Long-term Treasury index: Bond Fund (30 to 11-year)
This would also imply a preference for using the historical returns of Fidelity's short-term and intermediate-term index funds FUMBX and FUAMX.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic / international) stocks / (nominal / inflation-indexed) bonds | VCN/VXC/VLB/ZRR

User avatar
Kevin M
Posts: 10398
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by Kevin M » Wed Jan 16, 2019 8:11 pm

longinvest wrote:
Wed Jan 16, 2019 6:17 pm
My personal preference would be 1 to 5 years for short term and, correspondingly, 5 to 10 years for intermediate term. I find 1 to 3 years too narrow (only 2 rungs in the fund model). This choice would result into the following matches:
  • Short-term Treasury index: Bond Fund (5 to 2-year)
  • Intermediate-term Treasury index: Bond Fund (10 to 6-year)
  • Long-term Treasury index: Bond Fund (30 to 11-year)
This would also imply a preference for using the historical returns of Fidelity's short-term and intermediate-term index funds FUMBX and FUAMX.
Most of what you said makes sense on a quick read, but not so sure about this bit. On a forum called "Bogleheads", which is a descendant of a forum called "Vanguard Diehards", I would think we'd prefer the Vangaurd funds over the Fidelity funds. This also is more consistent with what Simba always has done, and I what Portfolio Visualizer does (use Vanguard funds when data available).

The VG short-term Treasury index fund holds bonds in the 1-3 year range, with an average maturity of two years. The intermediate-term fund holds bonds mostly in the 3-10 year range, with an average maturity of 5.6 years. Long-term Treasury holds 93% in 20-30 year range, with average maturity of about 25 years.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

Topic Author
longinvest
Posts: 3197
Joined: Sat Aug 11, 2012 8:44 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by longinvest » Wed Jan 16, 2019 8:45 pm

OK, it makes sense to stick with Vanguard's definitions*.

* Interestingly, Vanguard Canada defines short-term bonds (VSG) as as 1 to 5 years. Go figure!

How about the following?
  • Vanguard Short-Term Treasury Index Fund Admiral Shares (VSBSX):
    • "This low-cost index fund invests in debt issued by the U.S. Treasury (not including inflation-protected bonds) and U.S. government agencies, as well as corporate or dollar-denominated foreign debt guaranteed by the U.S. government, all with maturities between 1 and 3 years." (source)
    • Matching model: Bond Fund (3 to 2-year)
  • Vanguard Intermediate-Term Treasury Index Fund Admiral Shares (VSIGX):
    • "This low-cost index fund invests in debt issued by the U.S. Treasury (not including inflation-protected bonds) and U.S. government agencies, as well as corporate or dollar-denominated foreign debt guaranteed by the U.S. government, with maturities between 3 and 10 years." (source)
    • Matching model: Bond Fund (10 to 4-year)
  • Vanguard Long-Term Treasury Index Fund Admiral Shares (VLGSX):
    • "This low-cost index fund invests in debt issued by the U.S. Treasury (not including inflation-protected bonds) and U.S. government agencies, as well as corporate or dollar-denominated foreign debt guaranteed by the U.S. government, with maturities greater than 10 years." (source)
    • Matching model: Bond Fund (30 to 11-year)
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic / international) stocks / (nominal / inflation-indexed) bonds | VCN/VXC/VLB/ZRR

User avatar
siamond
Posts: 4410
Joined: Mon May 28, 2012 5:50 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by siamond » Thu Jan 17, 2019 1:24 pm

Understood and agreed about the various treasury index funds. Thanks, this is exactly the kind of guidelines that I was aiming at.

I share the feeling that Vanguard's approach to index IT/ST funds is a little weird, but it is what it is...
longinvest wrote:
Wed Jan 16, 2019 6:17 pm
3- Treasury funds (not indexed): when necessary, because of a lack of an index fund, use the model covering the same maturities as the fund.
Could you please elaborate, taking the example of the three active (and long lived) Vanguard funds I identified in this post (which are being tracked in Simba)?

Topic Author
longinvest
Posts: 3197
Joined: Sat Aug 11, 2012 8:44 am

Re: Historical Bond Returns - Shiller: From Rates to Returns [Bond Fund Simulator]

Post by longinvest » Thu Jan 17, 2019 6:23 pm

siamond wrote:
Thu Jan 17, 2019 1:24 pm
Could you please elaborate, taking the example of the three active (and long lived) Vanguard funds I identified in this post (which are being tracked in Simba)?
Sure. For active Treasury funds, I would use their benchmark as guide:

siamond wrote:
Tue Jan 15, 2019 11:14 am
Vanguard Short-Term Treasury Fund (VFISX/VFIRX):
- active fund, treasuries only
- benchmark: Bloomberg Barclays U.S. 1-5 Year Treasury Bond Index
Given the Bloomberg Barclays U.S. 1-5 Year Treasury Bond Index* benchmark, I would match it to Bond Fund (5 to 2-year).

* Includes U.S. Treasury obligations with maturities of 1 to 5 years.

siamond wrote:
Tue Jan 15, 2019 11:14 am
Vanguard Intermediate-Term Treasury Fund (VFITX/VFIUX):
- active fund, treasuries only
- benchmark: Bloomberg Barclays U.S. 5-10 Year Treasury Bond Index
Given the Bloomberg Barclays U.S. 5-10 Year Treasury Bond Index* benchmark, I would match it to Bond Fund (10 to 6-year).

* Includes U.S. Treasury obligations with maturities of 5 to 10 years.

siamond wrote:
Tue Jan 15, 2019 11:14 am
Vanguard Long-Term Treasury Fund (VUXTX/VUSUX):
- active fund, treasuries only
- benchmark: Bloomberg Barclays U.S. Long Treasury Bond
Given the Bloomberg Barclays U.S. Long Treasury Bond* benchmark, I would match it to Bond Fund (30 to 11-year).

* Includes U.S. Treasury obligations with remaining maturities of 10+ years.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic / international) stocks / (nominal / inflation-indexed) bonds | VCN/VXC/VLB/ZRR

Post Reply