"Bond fund returns=2 X duration minus 1"

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
jmk
Posts: 362
Joined: Tue Nov 01, 2011 7:48 pm

"Bond fund returns=2 X duration minus 1"

Post by jmk » Sun Aug 12, 2018 7:01 pm

How accurate is the rule of thumb "it takes 2 times duration minus 1 to earn the current yield on a bond fund"?

Or more accurately, “Current yield-to-worst of a bond fund/etf is a good predictor of annualized return over the next two times duration minus one years.”

For instance, we'd expect 11.2 years to earn 3.14% on TBM if rates rise?
Last edited by jmk on Mon Aug 13, 2018 4:12 pm, edited 3 times in total.

User avatar
vineviz
Posts: 1752
Joined: Tue May 15, 2018 1:55 pm

Re: "2 X duration minus 1"

Post by vineviz » Sun Aug 12, 2018 7:10 pm

jmk wrote:
Sun Aug 12, 2018 7:01 pm
How accurate is the rule of thumb "in rising rate environment it takes 2 times duration minus 1 to earn the current yield on a bond fund"?

For instance, might it take 11.2 years to earn 3.14% on TBM? Can it be much longer or shorter?
I’ve never heard of that rule of thumb and can’t think of why anyone woul espouse it.

Generally speaking, duration itself is a measure of the “break even” point on a bond.

But this 2x-1 thing seems baseless to me.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

User avatar
vineviz
Posts: 1752
Joined: Tue May 15, 2018 1:55 pm

Re: "2 X duration minus 1"

Post by vineviz » Sun Aug 12, 2018 7:27 pm

vineviz wrote:
Sun Aug 12, 2018 7:10 pm
jmk wrote:
Sun Aug 12, 2018 7:01 pm
How accurate is the rule of thumb "in rising rate environment it takes 2 times duration minus 1 to earn the current yield on a bond fund"?

For instance, might it take 11.2 years to earn 3.14% on TBM? Can it be much longer or shorter?
I’ve never heard of that rule of thumb and can’t think of why anyone woul espouse it.

Generally speaking, duration itself is a measure of the “break even” point on a bond.

But this 2x-1 thing seems baseless to me.
Hmm, I found the literature source for this idea. The number of implausible assumptions needed to make this kind of prediction work will limit its usefulness in portfolio planning I’m afraid.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

acegolfer
Posts: 1092
Joined: Tue Aug 25, 2009 9:40 am

Re: "2 X duration minus 1"

Post by acegolfer » Mon Aug 13, 2018 7:07 am

Here's a short academic answer.

A bond yield usually refers to yield to "maturity" (YTM). It's the annualized return, if one holds the bond till "maturity". OTOH, if he sells before the maturity, the return can be higher or lower than the YTM. In a rising rate environment, it's likely to be a lower return because the bond price decreases. So to earn the YTM, the bond holder may need to hold till maturity.

Next, what's 2 X duration minus 1?
Duration of a coupon bond is slightly larger than half of its maturity. OP's post suggests the bond maturity can be approximated by 2 X duration minus 1.

However, this explanation may not apply to a bond "fund" because it doesn't have a fixed maturity.

jmk
Posts: 362
Joined: Tue Nov 01, 2011 7:48 pm

Re: "2 X duration minus 1"

Post by jmk » Mon Aug 13, 2018 3:58 pm

The article that offers that rule of thumb, Corey Hoffstein, "How To Predict Bond ETF Returns":

https://www.etf.com/sections/etf-strate ... nopaging=1
To derive this rule, we make some simple assumptions.
•We are investing in a bond fund that seeks to maintain a constant duration by rolling over the portfolio at the end of each year. Most popular funds maintain a maturity target, not a duration target, but unless there are wild swings in interest rates, constant maturity is close enough to constant duration to make the rule apply.

•We estimate the return of our fund over a given one-year period to be the initial portfolio yield plus any price change due to changes in interest rates (so we ignore convexity effects, roll return, etc.)

•We assume bonds are trading at par, so we use “yield” and “yield-to-worst” interchangeably.
...
Let’s focus, for a moment, only on the returns from extra yield and losses due to rate changes. We can see, over time, how those two effects net out:
Image
If we hold for exactly “two times duration minus one” years—in this case, 9—the cumulative net excess profit and loss averages out to zero, and we end up just earning the starting yield every year.

Thus, starting yield becomes the predictor for return over the next “two times duration minus one” years.

Post Reply