Help me understand long-term bond funds (TLT)

[chrikenn]

Since August 31, 2010, TLT is down almost 18%.  Other treasury funds are down similar amounts.

Why?  Is it because treasury rates have been rising in that period?  Is it because stocks have done well in that period, so people are taking money out of treasuries and putting it into the stock market?

I'm just curious what's the primary driver behind the drop in NAV of treasury funds over the past 5 months.  Nearly 20% of value in a mere 5 months is quite a jump!

[moda0306]

On a very mathmatical level, it's because rates have risen.  The question then is, why have rates risen?  Probably a combination of higher expectations of economic recovery, higher expectations of inflation.

One thing that's interesting to look at is the treasury/corporate rate spread (don't know where to find this in a reliable, updated format).  It tells you how much interest rates represen all other risks other than default risk.  This spread went way up towards the end of 2008, and jumps like that are the reason that you see LT treasuries alone going gangbusters during that period, and why LT treasuries work so well for the PP.

[chrikenn]

Thanks for the replies.

I understand that NAV drops when interest rates rise, and that the amount of the drop is proportional to the duration of the bond.  I was mostly just curious if the entire drop over the last 5 months (or, the "primary driver" of the drop, as I called it) is attributable to interest rates, and I take it that it is (or at least, that it mostly is).

Just out of curiosity, say interest rates had remained completely flat since August 2010, but that stocks had continued to do well as they did.  We would expect that some people would sell bonds and go more into stocks, so perhaps there would be a small "flight" from treasuries into equities.  Would this also cause the NAV of TLT to drop?  i.e., is the NAV of a bond fund such as TLT determined entirely by interest rates, or does demand independently affect the NAV, too?

[MediumTex]

Interest rate movements are the sole determinant of the changes in NAV of TLT.

TLT pays out interest payments in the form of dividends, so there is no other factor to affect NAV except interest rates moves and their effect on the value of the underlying bonds.

The bond market takes a while to get the hang of, but once you begin to understand the relationship between interest rates, bond values, and the ways in which the stock market sentiment can affect the bond market, you begin seeing the inter-relationships among markets as part of a unified whole.

[moda0306]

Clive,

I don't think that spread is a default risk premium.  I think a spread between long and short government bonds simply represents peoples' short rate expectations.... they don't expect them to stay super low forever.

Those rate expectations are mostly made up of inflation and other economic factors, but I believe they don't really factor in much default since the government pretty much can't default unless they were to try.

[chrikenn]

Everyone has answered you correctly, but I think what you're trying to get at is does a flight-to-safety cause rates to go down and thus TLT to go up, and the answer is yes.

And why does flight-to-safety cause rates to go down?  Is it simply a function of demand---if more people want to buy treasuries, then the government can pay lower interest rates and people will buy them anyways?

I guess somehow I always thought that the interest rates paid by the government were not a function of demand for treasury bonds, but now I seem to be learning that they are indeed a function of demand.  Is that accurate?

Edit: Oh, and thanks to everybody for the replies.  This has been very helpful.

[AdamA]

"Since August 31, 2010, TLT is down almost 18%.  Other treasury funds are down similar amounts.
Why?"

I think are two main things that make Treasury bond yields go up (ie, price of fund go down).

1.  Optimism about other investments (the stock market, mostly).  If people think they can make more money in other markets, treasury bonds have to offer higher interest rates to compete.

2.  Fear of inflation.  If people think that the 5% APR they're gonna get from their 30 year bonds is going to get killed by inflation, they will obviously demand higher yields.


I think both of these things are occurring right now.

I think that's all you were asking...right?

Adam

[moda0306]

Though I would add a caveat to the market setting treasury rates would be that when the federal reserve purchases ST treasuries, it probably has some effect on LT rates based on changes in predictions in Long-term inflation and long-term recovery.

[Storm]

I guess somehow I always thought that the interest rates paid by the government were not a function of demand for treasury bonds, but now I seem to be learning that they are indeed a function of demand.  Is that accurate?

Right, the yield, or interest rate paid by the government, is set upon primary auction.  The government adjusts this up or down depending on demand and the market realities at the time of the auction.

After the bond has been sold in the primary market, the yield remains the same, however, the value of the bond fluctuates up and down depending on the future interest rate expectations of the market.  What we are seeing right now is that people expect interest rates to rise in the future, so the value of the bond goes down.

For a really simplistic example, let's say you bought a $1000 bond at primary auction with a 4% yield.  If the market expects interest rates to rise, they would pay less for your bond.  Maybe now you can only sell it for $900 because new bonds are coming out that have a 5% yield.  Market participants have the choice of buying a new bond for $1000 with a 5% yield or buying an older bond for $900 with a 4% yield.  They pay less up front because the bond will pay less income over time.

Please don't knock my numbers because I'm just giving a simplistic example to show the relationship between interest rates and value of the underlying bond.  I still don't know exactly how the value is calculated based on interest rate expectations, but hopefully this illustrates how interest rates and bond values can move inverse to each other.

[julian]

The prob with the 5 year is that in years like 2008 you would not had the huge gain of the TLT to offset the VTI loss

[chrikenn]

Just out of curiosity, then, is there any reason that Harry recommended the ST/LT barbell?  If a 50% intermediate term treasury performs nearly identically to a 25/25 ST/LT portfolio, it seems like the added simplicity (for those of us using funds, as opposed to holding actual bonds) would weigh in favor of the intermediate-term approach.  With that approach, I could have one less fund to deal with.

I get the impression that on this forum, most people stick with the 25/25 ST/LT barbell.  Other than the fact that this was Harry's recommendation, is there any actual benefit to going that route as opposed to 50% intermediate?

The only thing off the top of my head is that the 25/25 barbell would provide for more tax-loss harvesting opportunities.  That in and of itself could save tens of thousands of dollars over the course of an investing lifetime.  Other than that, though, any benefits to the ST/LT barbell as compared to 50% intermediate?

[Lone Wolf]

I get the impression that on this forum, most people stick with the 25/25 ST/LT barbell.  Other than the fact that this was Harry's recommendation, is there any actual benefit to going that route as opposed to 50% intermediate?

The issue here is that you limit your deflation protection.  If the US were to enter a prolonged deflationary period, very long-term bonds will provide you with a lot more juice than a 5-year.  For example, a Japanese PP suffers from the fact that there were no long-term (25-30 year) government bonds available to profit from collapsing interest rates.

Consider a scenario like the Great Depression where long-term interest rates fell all the way to 1%... and short-term T-bills were yielding negative interest rates!  If rates drop like that and stay there, LT bonds might have to save your entire portfolio.  5-years can't do it.

There are some other more minor issues like the fact that cash is more liquid than 5-year treasuries.  Clive does a good job of addressing this by constructing a ladder so there's still some good liquidity.  However, if I understand it correctly, this construct gives you even less deflation protection than holding all 5-years.  The reason is that a ladder provides you with an average duration more on the order of 2-3 years.  (If I have mischaracterized Clive's strategy, I hope that he will correct me.)

Clive, were the numbers that you used for a ladder or a PP with 5-year's?  Because I think that a ladder would have taken a worse beating in 2008 than what I saw in those numbers.

I think that Clive's tweak has shown good past results but I don't see any reason (now or in general) to deviate from the 4x25.  If the US were to ever stop offering LT bonds, I would adopt something much like what Clive proposes, though perhaps I'd go with a longer duration like 10 years rather than 5 years.  For now, though, LT bonds are there to defend against a protracted Japan-style (or worse) deflationary doldrums.

[MediumTex]

I believe that going 50% 10 year bonds instead of 25/25 ST/LT provides similar returns to the traditional HB PP.  However, in a severe and prolonged deflationary situation you would "grow out" of your protection with a 50% 10 year setup much more quickly than in a traditional 25/25 ST/LT configuration.

There is also the issue of not having anything to rebalance your treasury gains into if you are using a 50% 10 year approach.

[Storm]

Just out of curiosity, then, is there any reason that Harry recommended the ST/LT barbell?  If a 50% intermediate term treasury performs nearly identically to a 25/25 ST/LT portfolio, it seems like the added simplicity (for those of us using funds, as opposed to holding actual bonds) would weigh in favor of the intermediate-term approach.  With that approach, I could have one less fund to deal with.

I believe that the reason HB recommended the ST bonds as cash is because there is little to no inflationary risk.  With a 6 month time horizon, you simply buy new ones at a higher interest rate in 6 months.

In an inflationary period, the value of your LT bonds can go down significantly.  You don't have this risk with ST bonds.  In essence, buying LT bonds is sort of like a leveraged bet on future deflation.  Of course you can't lose more than your principal, so it's not the same as leverage.

HB would have wanted the cash portion to have no leverage, or no bets at all, since it is supposed to remain stable (in absolute value, not relative) in both deflationary and inflationary times.

[Lone Wolf]

I believe that going 50% 10 year bonds instead of 25/25 ST/LT provides similar returns to the traditional HB PP.  However, in a severe and prolonged deflationary situation you would "grow out" of your protection with a 50% 10 year setup much more quickly than in a traditional 25/25 ST/LT configuration.

I remember reading this at one point, too, but I don't remember how that was established.  Was this based on looking at Japan's deflation or some other data set?

[chrikenn]

Ok.  It makes sense to me that long-term would be better in prolonged deflationary periods.  Has the U.S. had any periods of prolonged deflation against which we can backtest?  None immediately come to mind.

Because, as has been pointed out, just as LT would be better in prolonged deflation, one would expect that intermediate-term would do slightly better (or should we say, slightly "less worse") in periods of prolonged inflation.  Using Simba's backtesting spreadsheet from Bogleheads for the years 1973-1981 (high inflation), the ST/LT barbell has the following statistics:

CAGR: 12.27%
Std. Dev.: 11.70

Using 50% intermediate-term instead of the ST/LT barbell yields the following:

CAGR: 12.21%
Std. Dev.: 11.75

That's not what I was expecting.  Given the prolonged inflation I figured int-term might have slightly outperformed.

In 2008 (closest thing to deflation I can get in Simba's spreadsheet), ST/LT barbell returned -0.71%.  50% Intermediate returned -1.35%.

At any rate, I guess I'm just thinking out loud and wondering if switching to 50% intermediates would be a way to simplify an already-simple portfolio.  It's probably not necessary.  Nobody can predict the future, but based on the backtesting I've done, it seems like over a 30-40 year investing lifetime, as Clive's chart suggests, 50% intermediates vs. ST/LT barbell is basically a wash.

[MediumTex]

Would the Mona Lisa look better if you put some hoop earrings on her?

I don't know, but since she has done so well without them I would be hesitant to change anything at this point.

[Lone Wolf]

Ok.  It makes sense to me that long-term would be better in prolonged deflationary periods.  Has the U.S. had any periods of prolonged deflation against which we can backtest?  None immediately come to mind.

Right.  The period that does come to mind (The Great Depression) had us on a Gold Standard.

It's also interesting to note that until 2008, we had no deflationary periods of any length to look at.  If you had sized up the PP before then, a backtest might have shown you that holding LT bonds at all was a waste of time.  Fortunately, when one came along, the Portfolio performed just like it was supposed to.

At any rate, I guess I'm just thinking out loud and wondering if switching to 50% intermediates would be a way to simplify an already-simple portfolio.  It's probably not necessary.  Nobody can predict the future, but based on the backtesting I've done, it seems like over a 30-40 year investing lifetime, as Clive's chart suggests, 50% intermediates vs. ST/LT barbell is basically a wash.

Could very well be true.  However, I am convinced that the 5-year approach isn't going to hold up as well under intense, prolonged deflationary pressure.  While we may not ever see such a situation, I'd just go with the instrument that we know would knock it out of the park (LT bonds.)

In all honesty, I'd expect Clive's version to do very well in the future.  However, I just feel much better when all my bases are covered.  If "it" (long bouts of deflation) happened in Japan, I can't rule out that it could happen here.  We have the good fortune of being able to hold LT bonds as protection, something the Japanese did not.

[moda0306]

To put some perspective on "long periods of deflation" we need to keep in mind that the rates of return on 30-year bonds take into consideration expected inflation... so though we haven't had a "long period of deflation" since 1982, the high inflation that were priced into the bonds (that didn't happen) has resulted in great returns for a LT treasury investor for the last 28 years.  It didn't take deflation, just prolonged disinflation.

So while a lot of people say, "why invest the same in a deflation asset as an inflation asset when inflation is so much more likely than deflation."  Answer: Expected inflation is priced into the nominal return of these LT bonds, and if inflation doesn't occur, or the expectation of inflation is lowered, then LT bonds will gain.  It doesn't take actual depression-esque deflation for LT bonds to perform.  Just a change downward in inflation realities and expectations.

[Lone Wolf]

Good point, Moda.  Big-time deflation is not the only time that LT bonds see some upswing.

However, when you do hit a hard deflationary period, LT bonds are probably going to be carrying the portfolio all by themselves while stocks and possibly gold crumble.  That is the one situation where nothing can replace LT bonds (at least ideally.)  It is when they become indispensable (at least IMO.)

[cowboyhat]

I like Clive's idea, but if you are working with a 401K account the 50% allocation of 5 year bond ladder is even harder to obtain than 25% long treasuries + 25% money market. I would need 3/4 of my assets outside my 401K.

[chrikenn]

Thanks for the discussion.  I have concluded I'll probably stick with the long-term bonds rather than going 50% intermediate.  Reasons being: 1) As pointed out above, we don't know exactly how things would react in a long period of deflation, but the long-term bonds are a safer bet than intermediate-term, and 2) even if we never see a period of prolonged deflation, having the ST/LT barbell gives me additional opportunities to tax-loss harvest  :D

[moda0306]

cowboyhat,

Good point.  Plus, Clive, your depression scenario used 20 year vs 25 or 30 year treasuries, and I assume t-bills are 1 year??... while I agree with Craig that 1-3 year fund is about as good as one can do with cash.

Maybe the combo of using 25 or 30 year treasuries and 1-3 year would tilt it slightly if not significantly back into the traditional PP's favor.

[Lone Wolf]

That is pretty cool info, Clive.  However, a few things.  You are measuring 10-year treasuries here, while you are recommending 5-year treasuries.  I think the difference could be pretty big.  (I like the sound more of the 10-year version myself, though I do worry about 1981 a bit.)

Also, like Moda points out, using 20 year treasuries vs 25 or 30's when measuring the PP is a skew in the wrong direction.  Browne has you selling them at 20 and buying them at 30.

Cowboy, excellent point on tax-deferred space being used more efficiently by the LT bonds as well.  I had not thought of that.  That's another mark in favor of LT bonds.

In addition, you are recommending a ladder.  This has some nice benefits but it's going to give you softer deflation protection.  Remember, this portion has to make up for suffering stocks and gold, so with bonds on average only 2.5 years from maturity you will have very little capital appreciation.  (If you go with the 10-year version, the average is 5 years, which is still less than I'd want.)

Don't get me wrong, I think that the 50% intermediate-term treasury approach is still a nice, robust allocation and will almost certainly do great going forward.  But I do believe that there are some (small) risks to laddering it.  In addition, I don't think 5-years have a big enough horizon on them.  Finally, having no cash or cash-like allocation (such as 50% in 10-year treasuries) is rough riding during a recession if the Fed slams the brakes on interest rates.

[MediumTex]

Also, as a practical matter, laddering takes away your ability to do PP "gadget plays" like using I-bonds for a portion of your cash holdings.