Doc wrote: ↑
Sun Apr 29, 2018 9:34 am
Kevin M wrote: ↑
Sat Apr 28, 2018 10:07 pm
Doc wrote: ↑
Sun Apr 29, 2018 9:34 am
And as I have mentioned before if the curve moves against you just hold to maturity. You don't "lose" the original return, you just don't get the "ride" bonus. You do however get a higher coupon until you sell because you took on more term risk.
Yeah, but in this case you'd probably earn more by rolling shorter maturities to the higher yields more quickly.
Hindsight is 20-20.
Huh? We're both talking about "what if" the yield curve moves against you--in other words, the shorter term yields increase, reducing or eliminating the return you expected due to the riding the yield curve effect. You say in that case you can just hold to maturity. I say in that case you could earn a higher return by rolling the shorter-term maturities into higher yields sooner. This isn't hindsight, this is just two different standard bond strategies, and we don't know in advance which will earn more.
My main point is just that riding the yield curve does not provide any sort of guaranteed extra return. It's just part of the potential, but not guaranteed, higher return you might
get for taking more term risk. So of course you don't lose the original YTM if you hold to maturity--it's just that that doesn't guarantee that you'll earn more than rolling shorter-term bonds over the same holding period.
This is an important point for people to understand, as I often see comments from posters about receiving a roll yield or roll return as if it's some guaranteed thing, but it's not.
I didn't say easier I said simpler. As you say it gets difficult to understand what "roll yield" really means. Is it only the capital component or does it include the coupon difference. I use the YTM (zero coupon) method because it includes both and YTM is the common way to look at bonds.
You are garbling terminology here. YTM applies to all bonds, whether coupon or zero coupon. We both are using YTM. You can either get your entire YTM from capital return with a zero, or from a combination of income return and capital return from a coupon bond. I'm worried that equating YTM to a zero coupon bond will confuse people (this conversation probably is already too deep in the weeds for most forum members).
Simplicity is in the eye of the spreadsheet creator. I find my method very simple and much more understandable than yours, but our brains work in different ways, so I understand that you find working with zero-coupon models easier. I also think my approach helps in understanding the difference between the income component and capital return component of bonds, and since most people own bonds or bond funds that pay income from coupons, I think this is useful.
For example, Vanguard provides a web page for each of their bond funds that separates the annual returns into income return and capital return, and what I'm doing is a way to help understand how this works at the individual bond level.
If you are using total return not just capital return the 7-5 is 7.71% and the 5-3 is 6.9% for a total return of 14.61% which maybe just perhaps(?) is exactly the same as the 7-3. (I'm using the Fed constant maturity data so I don't have the four data.)
We're both using the same constant maturity Treasury data (https://www.treasury.gov/resource-cente ... data=yield
). I just use linear interpolation to approximate the yields for the maturities they don't provide, which of course is not as sophisticated as the Fed/Treasury model, but good enough for the rough modeling we're doing. After all, the yields on Treasury.gov are just modeled yields, not yields you can get from a bond you can actually buy on that day.
Since my model is a ladder with rungs at each year, my modeled static yield curve returns will be slightly lower than yours, but it's not a big deal. Your 7-5 will be higher because you are using the 7-year initial yield for the 2-year period, while I am using the 7-year and interpolated 6-year yields, so the 7-6 part will be about the same, but the 6-5 part will be a bit lower in my model because the interpolated 6-year yield is lower than the 7-year yield.
So I get 6.39% total return for 7-5 and 6.11% for 5-3, for total of 12.49% for 7-3. This difference isn't particularly important for this discussion, but I'm glad that I understand it now.
Incidentally, the Treasury yields are based on bonds that are close to par, since they use the most recently auctioned Treasuries as much as possible, so the yields are more applicable to par bonds than to zero-coupon bonds. But again, for the type of rough modeling we're doing, it doesn't matter much.
I never said the you cannot do better with the CD. I said that the current "premium" after tax is not enough for me to make up for the lack of liquidity for the CD. I keep seeing and hearing about "no bids" for secondary CD's and early withdrawal penalties for direct CD's. That to me is not liquid.
OK, but the point I've been trying to make is that the "riding the yield curve" effect that you're hoping for may not materialize, and the relatively flat yield curve makes it even less likely. You're just not getting much bang for the buck in extending maturity, and yields have to rise much less to wipe out your ride return. You can get the same or more bang by buying a CD without having to extend maturity by a couple of years to get the higher yield.
We have no disagreement at all about the much higher liquidity of Treasuries compared to brokered CDs.
I would quibble with the early withdrawal penalty being considered a negative though, especially if it is low enough (which these days it typically isn't), since the early withdrawal penalty can be much less than the loss in a Treasury if yields increase enough. But that game isn't really on the table these days, since the yield premiums in direct CDs aren't high enough, and the EWPs generally are too high. That's why I've shifted to brokered CDs where the yield curve is steep enough that I feel sufficiently rewarded in taking the term risk.
Kevin M wrote: ↑
Sat Apr 28, 2018 10:07 pm
I get 3.04% vs. 2.71%, but close enough. Again, I wouldn't compare it this way, and this only applies if yield curve is static.
The difference is the coupon vs total return method I think. But yes it only counts if the yield curve is static.
First, I typed the wrong thing. I meant to compare my 3.04% to your 3.14% (not to your 2.71%), so it's even closer. And I think I've now found the main reason for the differences, as discussed above. We can both look at total return with our models, so there is no "coupon vs. total return" issue. In my model, total return = coupon return + capital return, while in your model total return = capital return. But there may be some additional minor differences in using zero-coupon vs. coupon bonds, but not enough to worry about.
I'm just looking for the "sweet spot" for a roll ladder not the sweet spot for something held to maturity.
I get that. I think we just have a different view of where the sweet spot is, but even more so how sweet it is.
I'll grant you that extending maturity might look slightly better in terms of bps/year if you add in a roll return component assuming a static yield curve. For example looking at extending from 6-year to 7-year (again, using interpolation for the 6-year) you get 6 bps looking at just yield (YTM), and you get 13 bps looking at total return (including the capital return component for par bonds assuming static yield curve). But even 13 bps just doesn't look very enticing to me.
By contrast, you get 25 bps of extra yield and 50 bps of extra total return (assuming static yield curve) by extending from 1-year to 2-year maturity.
If you're just looking for the peak of the expected total return assuming static yield curve, then yeah, 7-year looks the best (in the 1-10 year range). But if you're looking at it on a risk-adjusted basis--i.e., how much extra expected return you get for extending maturity by a year--then the 2-year looks best.
I do not try to maximize the the return from the fixed income portion of my portfolio. I'm not that smart. I can't predict FI future returns. I avoid long bonds because of the risk.
I avoid longer maturity bonds for the same reason--we just have a different definition of what "long" is. Long to me is anything that doesn't give me at least 20 bps of extra yield for an extra year of maturity. I'm just trying to optimize risk-adjusted return, not absolute return or absolute risk.
(You belive that the short part of the curve currently yields better returns.)
No, not better returns, better risk-adjusted expected returns. I have no idea what actually will produce higher returns, but we have to use some policy or strategy to make our decisions (and for many that is just to stick with a total bond fund, which is fine if that floats their boats).
I don't change my overall short term, long term position. I model my FI portfolio on the BBgBarc Intermediate (1-10) bond index. I break it into pieces for tax and equity correlation reasons into 4 parts. The treasury parts are short and intermediate. I use a roll ladder for the intermediate part and position it at the "best" area as shown in the chart above. (it's a pretty squishy area.) I then add enough short term Treasuries usually as funds to adjust the total maturity to match the index.
That's great. You have an investment policy and you're sticking to it.
My fixed-income investment policy is different, and it evolves as the fixed income landscape evolves. If my policy was to stick with direct 5-year CDs, I wouldn't be doing what I thought made sense now. It made a lot of sense a few years ago when I could get 100-150 bps yield premium over Treasuries of same maturities, and with EWPs of six months of interest. That just doesn't work anymore, so I've had to adapt.
Larry Swedroe isn't sticking with the same fixed-income strategy he had a few years ago. He's now into alternative fixed-income funds, but I'm not ready to go there. It just seems to be too much free lunch, not consistent with efficient markets. He's basically saying you get equity-like returns with fixed-income like risk. While I respect Larry immensely, I don't always go along with what he thinks makes the most sense for him. Similarly, I don't heavily tilt to small-value as he does (although I do tilt some).
Your take a maximize return approach within a credit risk criteria. You apparently also are willing to adjust the term risk to try to increase return. That's fine if that's what your IPS says.
Again, not trying to maximize return, but just to find the balance between expected return and risk that seems optimal to me. But yeah, perfectly willing to adjust term risk depending on the expected return, shape of yield curve, availability of great direct CD deals, etc.
I'm also willing to change the particular types of fixed income I use. A couple of years ago I wouldn't have considered any Treasuries, but now they look good out to 1-year maturity in tax-advantaged, compared to other choices. And Treasuries look better in taxable for me than CDs out to 3-year maturity, and about the same in terms of TEY at 5-year maturity. CDs have an edge of about 15 bps at 10-year maturity, but I wouldn't consider going that far out with anything with the yield curve so flat between 5-year and 10-year maturities.
I also wasn't considering individual munis until late last year, since then I have been buying them. Not only are the great direct CD deals not there any longer, so not good enough for taxable, but I decided to do my homework and learn about munis to the point where I felt comfortable with them. I refine my investment policies as circumstances change and as I learn more.