Vanguard Intermediate Term Treasury (VFITX) - Actively Managed

[raven15]

If you ever discover something amazing, the most likely explanation is you made a mistake.

I was comparing intermediate term treasury funds just now (exciting, I know) and what I found surprised me. As we know, Vanguard has an intermediate term treasury bond mutual fund, which has a low expense ratio despite being actively managed. I guess most people use it as an example of a well managed fund. I was therefore very surprised when I back tested it over the past year. It looks like Vanguard's fund was following Schwab's intermediate treasury ETF (SCHR) very closely until Christmas 2016, when the Vanguard fund managers made a bad interest rate bet, and the fund suddenly jumped below it's competitors.

VFITX is orange line


I looked back farther, and it has done even worse over the past five years, trailing all of its competitors. It seems the fund has made a series of bad interest rate bets. I have read that the bond market is notoriously hard to beat, but I am still surprised at the toll of active management.


Am I missing something? It seems like the Vanguard Intermediate Term Treasury mutual fund is something to be avoided in this space. The total treasury ETF GOVT is looking to be the gold standard, with SCHR or pretty much any indexed treasury fund being right up there. Which is funny because I see VFITX discussed more frequently than any of the others, I believe.

Source of My Innuendo

[saltycaper]

That's a price comparison chart. Try total return and it will look different. If you chart the mutual fund first, then compare the ETF, you'll see VFITX and SCHR are hardly distinguishable. (M* plots price if you start with an ETF, total return if you start with a mutual fund.)

[baw703916]

I hope Vanguard has a good method to identify those Treasuries that will outperform!

Seriously, the only effect of a lot of these bond funds being actively managed seems to be that the Admiral minimum is $50,000.

[stlutz]

On 12/22 VFITX had a capital gain distribution of about 1%.

[raven15]

Wow, price charts are super annoying. I thought simply adding VFITX would make it total return. I wish they would default to total return, or at least have big red text telling what it is and having a large and obvious button to toggle between. I figured someone would have said something by now, I should have slept on it before I posted .

Total Return Chart shows nearly identical

[Doc]

I hope Vanguard has a good method to identify those Treasuries that will outperform!

Seriously, the only effect of a lot of these bond funds being actively managed seems to be that the Admiral minimum is $50,000.

Why is your metric "outperform" and what does outperform even mean? Without looking I think that VFITX has had a shorter duration than it's index competition in the recent past. So are you looking at absolute returns or risk adjusted returns?

In addition the comparable Vg index fund is " Government" not Treasuries meaning it has agency investments which are often not "full faith and credit" obligations.

I'm not making a " better" judgement. Just saying it's not apples to apples.

[Doc]

On 12/22 VFITX had a capital gain distribution of about 1%.

Capital gains are taxed at a lower rate than interest. The cap gains is probably a result of roll yield which is usually thought of as a good thing.

[raven15]

On 12/22 VFITX had a capital gain distribution of about 1%.

Capital gains are taxed at a lower rate than interest. The cap gains is probably a result of roll yield which is usually thought of as a good thing.

The funny thing is, this started as an investigation of "roll yield." I was thinking that intermediate term funds should, on average, come out ahead of a total treasury fund such as GOVT because of the yield curve effect, despite having a similar duration. That actually looks to be the case, now that I am using the right chart .

[baw703916]

I hope Vanguard has a good method to identify those Treasuries that will outperform!

Seriously, the only effect of a lot of these bond funds being actively managed seems to be that the Admiral minimum is $50,000.

Why is your metric "outperform" and what does outperform even mean? Without looking I think that VFITX has had a shorter duration than it's index competition in the recent past. So are you looking at absolute returns or risk adjusted returns?

In addition the comparable Vg index fund is " Government" not Treasuries meaning it has agency investments which are often not "full faith and credit" obligations.

I'm not making a " better" judgement. Just saying it's not apples to apples.

I was being facetious. I can think of no good reason for active management for this fund, and yes the idea that some Treasuries will outperform is ridiculous.

But the higher Admiral threshold is real.

[stlutz]

Capital gains are taxed at a lower rate than interest. The cap gains is probably a result of roll yield which is usually thought of as a good thing.

Even better is an ETF that doesn't need to distribute capital gains--you can defer those until you sell instead of having to pay tax now.

One time I made the mistake of buying VFITX in taxable. Then I got a big *short-term* capital gain. Then I realized that Intermediate Term Government Bond (which has an ETF share class) was a better choice in taxable.

[saltycaper]

...

Without looking I think that VFITX has had a shorter duration than it's index competition in the recent past. So are you looking at absolute returns or risk adjusted returns?

In addition the comparable Vg index fund is " Government" not Treasuries meaning it has agency investments which are often not "full faith and credit" obligations.

I'm not making a " better" judgement. Just saying it's not apples to apples.

As an aside, Vanguard Intermediate-Term Government Bond Index Fund is something like 97% Treasuries, and duration (and performance) is nearly identical with Vanguard Intermediate-Term Treasury Fund.

[Doc]

As an aside, Vanguard Intermediate-Term Government Bond Index Fund is something like 97% Treasuries, and duration (and performance) is nearly identical with Vanguard Intermediate-Term Treasury Fund.

The duration of the active fund had been lower by ~1 year until the last few months IIRC so the risk adjusted return has not been nearly identical.

The $50k minimum however may be a deal breaker for many.

[saltycaper]

The duration of the active fund had been lower by ~1 year until the last few months IIRC so the risk adjusted return has not been nearly identical.

Not to be a PITA, but I really think it has. According to Portfolio Visualizer, for instance, a portfolio of 100% VSIGX has a Sharpe ratio of .92, versus .91 for VFITX, if measured from inception of VSIGX. That is nearly identical, IMO. More realistically, if you add either VSIGX or VFITX to a portfolio of, say, 40% or 50% US Total Stock Market and 20% or 30% Total International, the Sharpe ratios are exactly the same. Thus far, it really hasn't mattered which one of these funds you've owned, as far as performance and risk go. Any differences in risk-adjusted return between the two bond funds are so minor that they have been overwhelmed by swings in the stock portion of the portfolio. Even if we wanted to break down just the bond funds and be academic about it, we're still comparing apples to apples. Maybe Gala apples to Pink Lady apples, but certainly no oranges here.

[Kevin M]

The duration of the active fund had been lower by ~1 year until the last few months IIRC so the risk adjusted return has not been nearly identical.

Not to be a PITA, but I really think it has. According to Portfolio Visualizer, for instance, a portfolio of 100% VSIGX has a Sharpe ratio of .92, versus .91 for VFITX, if measured from inception of VSIGX.

If you use monthly returns instead of annual returns, you get Sharpe ratios of 0.75 for VFITX and 0.78 for VSIGX: Backtest Portfolio Asset Allocation. Not sure why so much change, since you only pick up one month of data, Dec 2009, by looking at monthly returns. At any rate, VSIGX still comes out slightly on top, but the difference is so small as to be essentially meaningless.

Max drawdown also is slightly greater for VFITX, at -4.30% compared to -3.98% for VSIGX.

So yeah, they look very similar on a risk adjusted basis.

Kevin

[Doc]

The duration of the active fund had been lower by ~1 year until the last few months IIRC so the risk adjusted return has not been nearly identical.

Not to be a PITA, but I really think it has. According to Portfolio Visualizer, for instance, a portfolio of 100% VSIGX has a Sharpe ratio of .92, versus .91 for VFITX, if measured from inception of VSIGX.

So yeah, they look very similar on a risk adjusted basis.

The Intermediate-Term Treasury Fund in particular was affected— its average duration at the end of the period stood at 5.3 years, compared with 6.6 years for its benchmark.

I was addressing term risk not variance. BUT I didn't recall correctly. The active fund is benchmark against BloomBarc US 5-10 Yr Treasury Index and the index fund against the BloomBarc US 3-10Yr Gov Flt Adj Idx. I miss-interpreted the "5.3 years, compared with 6.6 years" thinking both funds were bench marked against the same index. My bad.

I am in the process of considering breaking up the Vg intermediate Term Index into its components and didn't realize how similar the two Treasury/Government funds have been in the past. Thanks guys. I don't think I will use either one.

For what it's worth duration risk is more likely to raise its ugly head in the near future as (if?) intermediate term yields increase which seems to be the conventional wisdom. The timing of that increase is yet to be determined.

[alex_686]

I was being facetious. I can think of no good reason for active management for this fund, and yes the idea that some Treasuries will outperform is ridiculous.

I can think of a score. Most bond indexes only cover 10% of the most actively bonds, if you are willing to take a little liquidity risk you can dip into some of the less actively traded bonds. For example, "on the run" bonds are more expense than "off the run bonds". The index says you should buy a new 10 year treasury - but you can by a 9 1/2 one for cheaper. The index says to buy a new 5 year bond, but there is a slightly cheaper 10 year treasury which is 5 years old - so a maturity of 5 years. Now I am only talking about 10 to 20 bps but I will take free bps any day of the week.

[Kevin M]

I was addressing term risk not variance. <snip>

For what it's worth duration risk is more likely to raise its ugly head in the near future as (if?) intermediate term yields increase which seems to be the conventional wisdom.

Term risk should show up in the variance, and hence in the Sharpe ratio, which is a standard measure of risk-adjusted return. We can see an example of this by comparing the Vanguard short-term, intermediate-term and long-term Treasury funds: Backtest Portfolio Asset Allocation. As expected, SD increases as term risk increases, but not necessarily as expected, Sharpe ratio decreases as term risk increases. This supports the common mantra of keeping fixed-income duration short and taking risk on the equity side, and probably is one of the reasons for this notion.

But for the reason you state, I thought looking at max drawdown might be a better indicator. The period covered by PV includes 2013, a year in which term risk showed up to some extent, and the year with the largest drawdowns for VFITX and VSIGX since the inception of the latter. That was also the year the long-term Treasury fund had it's maximum drawdown of -17%!

Kevin

[Doc]

Most bond indexes only cover 10% of the most actively bonds, if you are willing to take a little liquidity risk you can dip into some of the less actively traded bonds.

I always thought of the liquidity issue backwards. The index fund has to buy the illiquid issue at high transaction cost. Meanwhile the active fund just buys something close that is more liquid. I assumed this was the reason that Vg went to the free float version of the BarCap Agg which seemed to exist for Vg only. Now that Bloomberg bought the index I think the distinction has disappeared.

[Doc]

Term risk should show up in the variance, and hence in the Sharpe ratio, which is a standard measure of risk-adjusted return.

Normally yes. But the bond market has not been normal since Lehman thus unfortunately, making risk measures like standard deviation somewhat suspect if we are only looking at the past 9 or 10 years. The term risk didn't show up because the yield curve was static.

I don't know the answer here. I already screwed up equating a 3-10 with a 5-10 by mistake. I just missed it.

As I tried to imply up thread, the active/passive question between well managed low cost funds like Vg's is in my mind less important than the bogey.

[alex_686]

I always thought of the liquidity issue backwards.

It is the other way around. Two big reasons are:

"Reconstituteablity" of the index: People constantly trade into and out of the fund. Funds want to easily buy and sell to match.

Pricing: Liquid securities trade frequently (by definition) so you have lots of good pricing data. Illiquid ones don't. If a bond hasn't traded for a week, which is common, how do you handle "stale pricing" - a industry term? There are ways, but they are not as good as actually trades in the open market.

[Kevin M]

Term risk should show up in the variance, and hence in the Sharpe ratio, which is a standard measure of risk-adjusted return.

Normally yes. But the bond market has not been normal since Lehman thus unfortunately, making risk measures like standard deviation somewhat suspect if we are only looking at the past 9 or 10 years. The term risk didn't show up because the yield curve was static.

I don't know why you're saying this. If you click the link to the PV comparison of short-, intermediate-, and long-term Treasury funds, and change the start date to January 2009 (after Lehman), you'll still see significantly higher standard deviation (and hence variance) for as the duration increases: SDs are 1.14%, 4.07% and 11.90% respectively. Sharpe ratios also decrease as term increases over the post-Lehman period.

If you don't like SD, look at max drawdown. I'd say a drawdown of -17% in 2013 for the long-term fund was the term risk showing up--this was a larger drawdown than anything before Lehman.

The yield curve most definitely has not been static since 2009. If it had been, we wouldn't see an SD of 11.90% for the long-term fund, and we wouldn't have seen a 17% decline for that fund in 2013 (and that's with monthly data, so probably worse for daily data). The term risk has shown up, but it also has been rewarded since 2009--in absolute terms, but not in risk-adjusted terms.

Interestingly, 2013 was only the fourth-worst drawdown for the intermediate-term fund, but it was of similar magnitude to the 2nd and 3rd-worst drawdowns, with the max drawdown at -6.47% occurring in 1994.

I don't see evidence that term risk for bonds, or how to measure it, has fundamentally changed since Lehman. Of course longer duration should be reflected in higher values of risk measures, and if that doesn't seem to be the case, I guess we have to look a little more deeply as to what's going on.

Kevin

[grok87]

i have never understood this fund. as i think other's have pointed out it seems to run with a duration a year or more less than its target index.
and the way interest rates have gone the past say 10 years that means it has underperformed it's index.

i am in the camp that there is no good reason for vanguard not to run this fund as an index fund.

[baw703916]

I was being facetious. I can think of no good reason for active management for this fund, and yes the idea that some Treasuries will outperform is ridiculous.

I can think of a score. Most bond indexes only cover 10% of the most actively bonds, if you are willing to take a little liquidity risk you can dip into some of the less actively traded bonds. For example, "on the run" bonds are more expense than "off the run bonds". The index says you should buy a new 10 year treasury - but you can by a 9 1/2 one for cheaper. The index says to buy a new 5 year bond, but there is a slightly cheaper 10 year treasury which is 5 years old - so a maturity of 5 years. Now I am only talking about 10 to 20 bps but I will take free bps any day of the week.

The liquidity issue does apply to munis and corporates, but not to Treasuries (including TIPS), which are by far the most liquid bonds in the world. Basically a Treasury index should contain all outstanding bonds with a certain range of maturity dates, in proportion to how many were issued. How hard is that to replicate? Plus, as with many small cap index funds, a reasonable of sampling is probably allowed.

[Doc]

I always thought of the liquidity issue backwards.

It is the other way around. Two big reasons are:

"Reconstituteablity" of the index: People constantly trade into and out of the fund. Funds want to easily buy and sell to match.

Pricing: Liquid securities trade frequently (by definition) so you have lots of good pricing data. Illiquid ones don't. If a bond hasn't traded for a week, which is common, how do you handle "stale pricing" - a industry term? There are ways, but they are not as good as actually trades in the open market.

That's exactly what a thought. If there is a net outflow from a fund the fund has to sell asset to raise the money. The index fund has to sell illiquid assets at a high cost as well as other holdings but the active fund can sell something else where the transaction cost is less. Am I missing something?

[grabiner]

I always thought of the liquidity issue backwards.

It is the other way around. Two big reasons are:

"Reconstituteablity" of the index: People constantly trade into and out of the fund. Funds want to easily buy and sell to match.

Pricing: Liquid securities trade frequently (by definition) so you have lots of good pricing data. Illiquid ones don't. If a bond hasn't traded for a week, which is common, how do you handle "stale pricing" - a industry term? There are ways, but they are not as good as actually trades in the open market.

That's exactly what a thought. If there is a net outflow from a fund the fund has to sell asset to raise the money. The index fund has to sell illiquid assets at a high cost as well as other holdings but the active fund can sell something else where the transaction cost is less. Am I missing something?

Index funds can use options and futures to track the index while maintaining some liquid assets; active funds can similarly use cash. ETFs avoid the issue entirely because they are only redeemable in-kind, but ETFs which hold illiquid assets tend to have larger spreads because traders cannot trust the pricing information.

In addition, if index funds might require sales of illiquid assets, Vanguard can charge a redemption fee so that investors who do not sell are not affected. Currently, the only Vanguard fund which has a redemption fee is Vanguard Global Ex-US Real Estate. (For most funds, Vanguard expects purchases to exceed redemptions, and thus it is more likely to charge a purchase fee than a redemption fee.)

[Doc]

I don't see evidence that term risk for bonds, or how to measure it, has fundamentally changed since Lehman. Of course longer duration should be reflected in higher values of risk measures, and if that doesn't seem to be the case, I guess we have to look a little more deeply as to what's going on.

The term risk hasn't changed it has just not come home to roost because the yield curve hasn't changed much in the last 9 years because of Fed action of keeping rates near zero. A 100% increase in a zero yield has little effect on the price. I'm just trying to point out that the last 9 years are atypical.

[Doc]

The liquidity issue does apply to munis and corporates, but not to Treasuries (including TIPS), which are by far the most liquid bonds in the world.

Two weeks ago it took me 3 days to sell a $40k position in a five year TIPS. The reason from the bond desk was low liquidity. There were no bids available at 2 of our 3 brokers.

[Theoretical]

Wow. That's crazy for a treasury security, especially with 40 bonds.

[raven15]

Anyone can do the research in just a few minutes on portfolio visualizer, but what I found was that I prefer these funds in approximately this order based on history since 2010 ish (many are relatively new):

VGIT/VSIGX Vanguard Intermediate Government: This fund includes agency debt, which increases yield a tiny amount, about 0.1% over SCHR. In theory there may be credit risk to agency debt, but it has not shown up and standard deviation has been identical to SCHR.

SCHR Schwab Intermediate treasury: Almost identical to VGIT, but fractionally less return because it contains only treasury bonds.

VFITX Namesake of the topic. Almost identical to both of the above, but worse. Fractionally lower return than SCHR, but fractionally higher standard deviation. May distribute relatively large gains, per stlutz. Do not choose this if you can choose either of the above.

The above three are almost identical in return and behaviour. The ones below are different takes on intermediate treasury funds.

GOVT iShares total treasury index fund: Slightly higher duration gives slightly higher volatility, but it does not benefit as much from "roll yield" which means it has had lower return despite the greater volatility. There may be an advantage to the "total treasury" structure, but for the most part it has not shown up so far.

IEF iShares 10-7 year treasury: obviously has a longer duration, which has resulted in higher standard deviation and so far mostly higher return. I don't see the advantage of splitting the 3-10 year into two segments, so I don't see the point here.

IEF iShares 7-3 year treasury: obviously has a shorter duration, which has resulted in lower standard deviation and so far mostly lower return. I don't see the advantage of splitting the 3-10 year into two segments, so I don't see the point here. I would not know which to chose.

[Doc]

Wow. That's crazy for a treasury security.

Negative yield and pp IT maybe is my guess but I was told otherwise.

Vanguard, the new kid on the street, was the only one with bids. Go figure.

[raven15]

I was being facetious. I can think of no good reason for active management for this fund, and yes the idea that some Treasuries will outperform is ridiculous.

I can think of a score. Most bond indexes only cover 10% of the most actively bonds, if you are willing to take a little liquidity risk you can dip into some of the less actively traded bonds. For example, "on the run" bonds are more expense than "off the run bonds". The index says you should buy a new 10 year treasury - but you can by a 9 1/2 one for cheaper. The index says to buy a new 5 year bond, but there is a slightly cheaper 10 year treasury which is 5 years old - so a maturity of 5 years. Now I am only talking about 10 to 20 bps but I will take free bps any day of the week.

So far it hasn't helped, VFITX has had very slightly less return and greater volatility than SCHR and VGIT, to which it is otherwise nearly identical.

[raven15]

...The active fund [VFITX] is benchmark against BloomBarc US 5-10 Yr Treasury Index .

I am in the process of considering breaking up the Vg intermediate Term Index into its components and didn't realize how similar the two Treasury/Government funds have been in the past. Thanks guys. I don't think I will use either one.

The 5-10 year benchmark doesn't make sense. It seems pretty clear in backtesting and on Vanguard's website that the fund is shadowing a 3-10 year index.

What components would you use and what advantage do you hope to gain?

[Doc]

...The active fund [VFITX] is benchmark against BloomBarc US 5-10 Yr Treasury Index .

I am in the process of considering breaking up the Vg intermediate Term Index into its components and didn't realize how similar the two Treasury/Government funds have been in the past. Thanks guys. I don't think I will use either one.

The 5-10 year benchmark doesn't make sense. It seems pretty clear in backtesting and on Vanguard's website that the fund is shadowing a 3-10 year index.

What components would you use and what advantage do you hope to gain?

It makes absolute sense. It's an active fund. Given the shape of the yield curve from 2008 until just recently the sweet spot on the yield curve has been in the 3-7 range when both term risk and roll yield are considered. That doesn't mean that they avoid the 7-10 part completely. They do what they think best not what some index dictates. Even Jack Bogle has been critical of the AGG index in recent years.

I'm using a 3-7 ETF and ~3-7 mixed TIPS/nominal ladder. (I also use some 1-3 maturity bonds or funds as short term reserves.) My overall bond portfolio is modeled on the 1-10 Gov/credit index but I mostly separate Corp & T's for tax efficiency reasons and to have a bigger gun when the next equity bear shows it's ugly head.

[Kevin M]

I don't see evidence that term risk for bonds, or how to measure it, has fundamentally changed since Lehman. Of course longer duration should be reflected in higher values of risk measures, and if that doesn't seem to be the case, I guess we have to look a little more deeply as to what's going on.

The term risk hasn't changed it has just not come home to roost because the yield curve hasn't changed much in the last 9 years because of Fed action of keeping rates near zero. A 100% increase in a zero yield has little effect on the price. I'm just trying to point out that the last 9 years are atypical.

It depends what you mean by "has just not come home to roost". If you mean that we haven't seen a negative nominal return for intermediate-term Treasuries for the 9-year period, then sure, but that's not unusual. Even if you look at simulated intermediate-term Treasury returns since 1871, there has been no 9-year period with a negative nominal return, with the lowest being 1951-1959 with an annualized 1.4% nominal.

If you look at 9-year periods since 1981, during which interest rates generally declined, the 9-year period 2008-2016 (inclusive) actually had the lowest 9-year nominal return for intermediate-term Treasuries at 3.8% annualized. The highest 9-year nominal return was 13.2% for 1982-1990. Neither of these results is particularly surprising, since they are in the ballpark of the initial 10-year yields at the beginning of the periods. Again, nothing particularly unusual about the last nine years.

The Fed has not kept "rates" near zero over the last nine years. They have kept the federal funds rate (FFR) near 0%, but that has had little relationship to Treasury yields in the 3-10 year range.

So I still think you're mischaracterizing the last nine years as something particularly unique for 3-10 year Treasuries. Treasury yields have been generally declining since 1981, with lots of up and down action along the way, and that trend simply continued over the last nine years. So looking at any holding period longer than one or two years since 1981, we could say that term risk has not come home to roost since 1981, if we define coming home to roost as a negative nominal return.

And looking at one-year periods, we've already seen that 1994 was the worst year for intermediate-term Treasuries since inception of the VG int-term fund, so before Lehman, while the worst year for long-term Treasuries was 2013, so after Lehman. Even looking at 1-year periods since 1871, 2013 was the worst 1-year return for long-term Treasuries at -13.0%, so the risk showed up more for long-term Treasuries in the post-Lehman period than it ever did before (again, everything here is in nominal terms). For intermediate-term Treasuries, 2008 was the fourth-worst one-year return since 1871, so nothing unique about post-Lehman. So the term risk has shown up for one-year periods both before and after Lehman. Same as always.

Saying that the yield curve hasn't changed much since Lehman also is a mischaracterization. Are you looking just at the yield curve at the beginning and end of the period, or at the yield curve changes along the way? Either way it's inaccurate to say that the yield curve hasn't changed much.

Before looking at that, let's get calibrated by zooming in to the period surrounding the Lehman failure.

Lehman failed on September 15, 2008. So let's look at the 10-year Treasury yield for September 2008.



We see that on 9/12/2008 it was 3.74%, then it fell to 3.41% by 9/17 (two days after Lehman), then it rebounded to 3.78% by 9/19 (four days after Lehman). So if we had gone on an unconnected vacation for a week, we wouldn't have detected anything unusual based only on the 10-year yields when we left and when we returned. It doesn't seem like the Lehman failure is anything in particular to anchor on.

Now let's zoom out and look at the 10-year yield since 1981, during which yields have generally been declining.


Do you see anything particularly unusual about the last nine years here? I don't.

Now let's add the 3-year Treasury to get a sense of yield curve changes since 1981.



Hmm. We see the yield curve flattening going into recessions, and steepening during and coming out of recessions. Again, I don't see anything particularly unusual about the last nine years.

To see this more clearly, let's look at the 10 minus 3 year yield spread since 1981:



Looks to me like pretty typical yield curve volatility over the last nine years. I certainly don't see a static yield curve over the last nine years. What do you see?

Kevin

[Doc]

It depends what you mean by "has just not come home to roost". If you mean that we haven't seen a negative nominal return for intermediate-term Treasuries for the 9-year period, then sure, but that's not unusual.

No I mean that intermediate term Treasury yields have not gone up substantially over the past several years even though many if not most think that these rates will rise. That term risk has not come home to roost. In fact there is one poster on this board who promotes using CD's with low early withdrawal penalties as one of the reasons to prefer CD's over Treasuries if/when interest rates rise. Oh gee whiz, that poster is you isn't it.

So are you arguing with me or with Kevin.

Using variance as a risk measure in a stable yield curve environment may be a useful metric even for Treasuries. But it ignores term risk in a changing yield curve environment unless the time period you are using covers many yield curve cycles.

Back in the days when the Sharpe ratio was widely discussed people often took total bond market (10 yr Treasury ?) as the riskless asset when the original CAPM model was I believe, 4 wk T-bills. Now we are trying to apply the Sharpe ratio to intermediate term Treasuries. I have trouble with that application.

My original position that the active and index funds under discussion did not having the same risk was based on my misconception that they had significantly different durations. I was wrong. I misinterpreted a comment in some report because I thought they used the same bogey which they don't.

There is no issue here as far as I am concerned.

[stlutz]

VGIT/VSIGX Vanguard Intermediate Government: This fund includes agency debt, which increases yield a tiny amount, about 0.1% over SCHR. In theory there may be credit risk to agency debt, but it has not shown up and standard deviation has been identical to SCHR.

SCHR Schwab Intermediate treasury: Almost identical to VGIT, but fractionally less return because it contains only treasury bonds.

The extent to which VGIT is not a treasury fund is commonly over-estimated. In fact, the fund is 97% Treasuries and only 3% other types of government bonds.

Right now the SEC Yield of SCHR is 1.77% while VGIT is 1.76%--a difference which corresponds exactly to the difference in their ERs.

Note that VGIT has pretty much paid small capital gain distributions every year which SCHR has not. That is the disadvantage to Vanguard's hybrid ETF/mutual fund structure.

[stlutz]

i have never understood this fund. as i think other's have pointed out it seems to run with a duration a year or more less than its target index.
and the way interest rates have gone the past say 10 years that means it has underperformed it's index.

i am in the camp that there is no good reason for vanguard not to run this fund as an index fund.

A couple of thoughts on this one. Although it's not explicitly stated anywhere, VG seems to target a duration in the low 5s for this fund, always. Anytime I've looked at this fund over the past 15 or so years, it's duration has been in the low 5s.

Duration of course is lower for a given maturity when rates are higher. So, right now this fund has an average maturity of 5.7, a yield of 1.6% and a duration of 5.2. If I pull up the annual report from 2005 for this fund, it had an SEC Yield of 4.4%, an average maturity of 6.9, and a duration of, you guessed it, 5.2.

Back then, the 5-10 index made a lot more sense as a bogey. But this also shows that running it as an index fund pegged to a 3-10 index (like VGIT) actually produces something different from what they have been doing with active management.

I'm sure nowadays you can find a "smart beta" index that targets an average duration, but those haven't been around for most of this fund's history.

[raven15]

VGIT/VSIGX Vanguard Intermediate Government: This fund includes agency debt, which increases yield a tiny amount, about 0.1% over SCHR. In theory there may be credit risk to agency debt, but it has not shown up and standard deviation has been identical to SCHR.

SCHR Schwab Intermediate treasury: Almost identical to VGIT, but fractionally less return because it contains only treasury bonds.

The extent to which VGIT is not a treasury fund is commonly over-estimated. In fact, the fund is 97% Treasuries and only 3% other types of government bonds.

Right now the SEC Yield of SCHR is 1.77% while VGIT is 1.76%--a difference which corresponds exactly to the difference in their ERs.

Note that VGIT has pretty much paid small capital gain distributions every year which SCHR has not. That is the disadvantage to Vanguard's hybrid ETF/mutual fund structure.

Odd. In practice it looks like VGIT has usually outperformed SCHR by about 5-10 basis points. I figured the explanation is either due to agency debt or Schwab making a profit. I'd think agency debt must be at least a part of it.

[stlutz]

Odd. In practice it looks like VGIT has usually outperformed SCHR by about 5-10 basis points. I figured the explanation is either due to agency debt or Schwab making a profit. I'd think agency debt must be at least a part of it.

Comparing the funds' benchmarks:

VGIT:

1 yr. .32
3 yr. 2.19
5 yr. 1.47

SCHR

1 yr. .31
3 yr. 2.19
5 yr. 1.43

A very slight lead for VGIT--had never noticed that before!

[grabiner]

Note that VGIT has pretty much paid small capital gain distributions every year which SCHR has not. That is the disadvantage to Vanguard's hybrid ETF/mutual fund structure.

For a taxable bond fund, the capital gain is an advantage, not a disadvantage; when rates rise and the fund sells bonds, you pay capital gains now instead of receiving larger dividends later which are taxed at a higher rate.

For a municipal-bond fund, capital gains are a clear disadvantage, but VTEB doesn't have a long enough history to determine what it will do with capital gains.

[Doc]

Odd. In practice it looks like VGIT has usually outperformed SCHR by about 5-10 basis points. I figured the explanation is either due to agency debt or Schwab making a profit. I'd think agency debt must be at least a part of it.

Comparing the funds' benchmarks:

VGIT:

1 yr. .32
3 yr. 2.19
5 yr. 1.47

SCHR

1 yr. .31
3 yr. 2.19
5 yr. 1.43

A very slight lead for VGIT--had never noticed that before!

Total return growth of $10k 11/01/2010 to date VGIT 11,628.93, SCHR 11,621.97 --- $8 on $10k for 7 years - that's one Starbucks every few years.

[Doc]

A couple of thoughts on this one. Although it's not explicitly stated anywhere, VG seems to target a duration in the low 5s for this fund, always. Anytime I've looked at this fund over the past 15 or so years, it's duration has been in the low 5s.

Is it the target duration that is the primary objective or is it positioning the portfolio on that part of the yield curve that produces the most roll yield?

David's comment:

For a taxable bond fund, the capital gain is an advantage, not a disadvantage ...

suggests that maybe it is the roll yield effect which also converts some of the interest payment into LTCG.

[alex_686]

O.k., lots of things to cover, so I am combining 2 different posts.

I always thought of the liquidity issue backwards.

It is the other way around. Two big reasons are:

"Reconstituteablity" of the index: People constantly trade into and out of the fund. Funds want to easily buy and sell to match.

Pricing: Liquid securities trade frequently (by definition) so you have lots of good pricing data. Illiquid ones don't. If a bond hasn't traded for a week, which is common, how do you handle "stale pricing" - a industry term? There are ways, but they are not as good as actually trades in the open market.

That's exactly what a thought. If there is a net outflow from a fund the fund has to sell asset to raise the money. The index fund has to sell illiquid assets at a high cost as well as other holdings but the active fund can sell something else where the transaction cost is less. Am I missing something?

Yes. There are 2 different types of indexes out there. One is "Tracking", which tries to figure out the market, but you can't invest in easily. Think real estate, hedge funds, private equity, etc. The other are "Investable" indexes, which is what we base our passive index funds on. The index providers want to sell these indexes to mutual funds, and mutual funds want to keep costs low. Not just the explicit "expense" ratio but the implicit costs as well, such as the bid / ask spread. Illiquid securities have big spreads, which is inefficient. So index providers do not include illiquid securities in their index. For example, the index that the Vanguard Total Bond fund uses only covers the 10% of the most liquid part of the market it is trying to track. Look at the methodology behind Bloomberg's Intermediate Term Treasury index and you will find they routinely throw out intermediate term treasuries because they can't get good pricing data on them.

I was being facetious. I can think of no good reason for active management for this fund, and yes the idea that some Treasuries will outperform is ridiculous.

I can think of a score. Most bond indexes only cover 10% of the most actively bonds, if you are willing to take a little liquidity risk you can dip into some of the less actively traded bonds. For example, "on the run" bonds are more expense than "off the run bonds". The index says you should buy a new 10 year treasury - but you can by a 9 1/2 one for cheaper. The index says to buy a new 5 year bond, but there is a slightly cheaper 10 year treasury which is 5 years old - so a maturity of 5 years. Now I am only talking about 10 to 20 bps but I will take free bps any day of the week.

The liquidity issue does apply to munis and corporates, but not to Treasuries (including TIPS), which are by far the most liquid bonds in the world. Basically a Treasury index should contain all outstanding bonds with a certain range of maturity dates, in proportion to how many were issued. How hard is that to replicate? Plus, as with many small cap index funds, a reasonable of sampling is probably allowed.

Index funds can use options and futures to track the index while maintaining some liquid assets; active funds can similarly use cash.

Well, active and passive can both use the same bag of tricks, and this works for many equity funds. The problem is that there are very few futures or options that track bond indexes.

ETFs avoid the issue entirely because they are only redeemable in-kind, but ETFs which hold illiquid assets tend to have larger spreads because traders cannot trust the pricing information.

I would disagree with you here, but it just may be a point of semantics. ETFs don't avoid the issue, they shift the issue to the Unit Creators, and the Unit Creators charge the ETFs via a large bid / ask spread.

The liquidity issue does apply to munis and corporates, but not to Treasuries (including TIPS), which are by far the most liquid bonds in the world.

This is both true and meaningless. "Off the run" Treasuries are illiquid when compared to "On the run" Treasuries. Try constructing a yield curve with both - you don't get a curve. You get big dips every time you hit the "On the Run" Treasuries - this liquidity premium is a real measurable thing. TIPS are illiquid when compared to "Off the run" Treasuries or UK inflation linked Gilts.

Basically a Treasury index should contain all outstanding bonds with a certain range of maturity dates, in proportion to how many were issued. How hard is that to replicate? Plus, as with many small cap index funds, a reasonable of sampling is probably allowed.

I am not sure if you are arguing for a complete universe or a sampling.

If you are arguing for a complete universe, can you point me towards any indexes that try doing this? Having worked on this issue I can tell you it is incredible hard. About 20% of the Treasury universe has stale prices - that is they have not been traded in the past 5 days. Another 20% have prices that are problematic. One can get around this by bootstrapping data but that leads to larger estimation errors for your index. All indexes keep a internal number on how big a estimate error their index have and try to keep it low.

If you are arguing for a reasonable sampling, that is what the industry does. Sample the most liquid 10% of the market because that is where the high quality pricing information is. Ignore the other 90%.

[Doc]

@alex_686

You make some good points.

One in particular:

Look at the methodology behind Bloomberg's Intermediate Term Treasury index and you will find they routinely throw out intermediate term treasuries because they can't get good pricing data on them.

I was aware that when Bloomberg took over Barclays "analytics" that they change the AGG index to a float adjusted basis (Sorry Vanguard you are no longer unique) I was unaware of the extent of the change on other indexes. Something to think about.

[Kevin M]

Since this thread had not seen fit to die yet ...

It depends what you mean by "has just not come home to roost". If you mean that we haven't seen a negative nominal return for intermediate-term Treasuries for the 9-year period, then sure, but that's not unusual.

Since this is the only part of my last reply you responded to, can I assume that you've accepted the refutation of your other points, such as the yield curve having been static over the last nine years, and "rates" having been held at 0% by the Fed?

No I mean that intermediate term Treasury yields have not gone up substantially over the past several years even though many if not most think that these rates will rise. That term risk has not come home to roost.

OK, but this is approximately the same thing, and again, is not unique to the last nine years compared to the last 35 years. What I have a problem with is your characterization of the last nine years as somehow unique in a way that standard measures of risk and risk-adjusted returns don't apply for some reason.

We saw term risk show up in 2013, and the drawdowns were proportional to duration (term risk), as we'd expect. Whether you look at drawdowns or standard deviation of annual returns, the performance of bonds over the last nine years give us just as much of a clue as to a potential distribution of future returns as any other 9-year period.

In fact there is one poster on this board who promotes using CD's with low early withdrawal penalties as one of the reasons to prefer CD's over Treasuries if/when interest rates rise. Oh gee whiz, that poster is you isn't it.

So are you arguing with me or with Kevin.

Perhaps you can be more specific, but I don't see any inconsistencies. There are two reasons I have preferred direct CDs to bonds in recent years (although I continue to hold some bonds): 1) a yield premium that has averaged 115 basis points over Treasuries of same maturities over the last 6.5 years, and closer to 150 basis points over the last two years; 2) the early withdrawal options with low early withdrawal penalties (EWPs) that significantly reduces term risk compared to Treasuries of same maturity. How is that inconsistent with arguing that term risk basics for bonds have not been unique over the last nine years compared to any other time?

Using variance as a risk measure in a stable yield curve environment may be a useful metric even for Treasuries. But it ignores term risk in a changing yield curve environment unless the time period you are using covers many yield curve cycles.

Now I'm confused. I thought you were arguing that variance and Sharpe ratio were not applicable to the last nine years because the yield curve has been static (which is not the case), but now you're arguing that variance is a useful risk metric in a stable yield curve environment?

At any rate, I agree that variance (or standard deviation) of annual returns is not directly applicable to holding periods of longer than one year. However, both theory and empirical evidence support the notion that long-term variance is proportional to shorter-term variance, so SD of one year returns should provide us with a sense of the distribution characteristics of expected returns for longer periods. This is at least the case for a bond fund or rolling ladder of bonds, with no maturity date.

So whether we look at SD of annual returns, maximum drawdowns during periods of rising interest rates as we saw in 2013, or simply average duration, we should come to the conclusion that longer-term bonds are riskier than shorter-term bonds, and that funds with no credit risk and the about same duration should perform about the same. And that's exactly what we see for the two funds the comparison of which started this line of discussion. I come to the same conclusion whether I look at at last nine years or the last 100 years.

Back in the days when the Sharpe ratio was widely discussed people often took total bond market (10 yr Treasury ?) as the riskless asset when the original CAPM model was I believe, 4 wk T-bills. Now we are trying to apply the Sharpe ratio to intermediate term Treasuries. I have trouble with that application.

I can't see any justification for using a rolling bond ladder or bond fund with a duration of five or more years as a riskless asset in any circumstances. A 10-year Treasury is the riskless asset in nominal terms for a 10-year holding period. A 1-year Tbill is the riskless asset for a 1-year period, and a 1-month Tbill is the riskless asset for a 1-month period (all in nominal terms of course).

Of course it would not make sense to calculate Sharpe ratio for a rolling ladder of intermediate-term Treasuries if a rolling ladder of intermediate-term Treasuries was considered to be the riskless asset, but it's not.

Most of the applications of CAPM I've seen have an implicit holding period of one year, since it's typically annual returns that are used in the analysis. This is what we seen in Portfolio Visualizer. And of course the Sharpe ratios are based on one-year returns and SD of one-year returns. Again, I agree that Sharpe ratios based on one-year returns aren't directly applicable to 5-year or 10-year or 30-year returns, but again, there ought to be some proportionality there.

If I want nominal certainty for a 5-year holding period, I'll prefer a 5-year Treasury or CD held to maturity over a bond fund with an average maturity of five years, since either of the individual securities is riskless in nominal terms for a 5-year holding period. Of course I'll prefer the CD to the Treasury if I can earn a decent yield premium.

Of course I'd prefer a 5-year TIPS if I want 5-year certainty in real terms, but with a yield premium over nominal Treasury of 100 basis points, I'm willing to live with the additional uncertainty for this relatively short time period. And the early withdrawal option of the CD provides an additional hedge against unexpected inflation, since nominal rates are likely to increase with inflation.

Back to the main point, I think PV uses 1-month Treasury as the riskless asset. It could be argued that 1-year Treasury is more applicable, since all the other components of the Sharpe ratio calculation are based on annual returns, but this is a relatively minor point. Whether you subtract the risk-free rate from the asset return in the numerator or not, you basically have a ratio of return to SD of returns. I don't think the fact that the risk-free rate has been close to 0% for the last nine years has huge significance. We see a similar pattern of Sharpe ratios increasing with duration for 1978-2008 as we do for 2009-2016.

My original position that the active and index funds under discussion did not having the same risk was based on my misconception that they had significantly different durations. I was wrong. I misinterpreted a comment in some report because I thought they used the same bogey which they don't.

It's good you figured that out, since it resolves what otherwise would be a bit of a puzzle, and we might be having an even more interesting discussion.

There is no issue here as far as I am concerned.

Not if we agree that there's nothing unique about the last nine years that should cause us to abandon well-understood risk and expected return measures for other than the shortest-term bonds (for which the Fed has not allowed an expected inflation component to be built into the yields).

Kevin