Initial SEC yield and subsequent 5yr and 10yr returns

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Topic Author
Kevin M
Posts: 11183
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Initial SEC yield and subsequent 5yr and 10yr returns

Post by Kevin M » Sun Jan 18, 2015 1:12 am

There have been many posts related to this topic, but I wanted to start a thread to show actual results for various Vanguard bond funds.

In this first post is a chart of initial SEC yields and subsequent 5yr and 10yr returns for Vanguard California Intermediate-Term Tax-Exempt bond fund (one of the funds I own). I used investor shares (VCAIX) because there is more history for it than for admiral shares. Returns for 2005-2014 are from Vanguard, returns prior to that are based on adjusted-close prices from Yahoo Finance, which are not perfect but are pretty close (most returns I checked against Vanguard for 2005-2014 are within a couple of basis points).

Image

Anyone want to make any observations?

Kevin
Last edited by Kevin M on Mon Jul 23, 2018 12:49 pm, edited 1 time in total.
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

dkturner
Posts: 1509
Joined: Sun Feb 25, 2007 7:58 pm

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by dkturner » Sun Jan 18, 2015 7:37 am

Kevin M wrote:There have been many posts related to this topic, but I wanted to start a thread to show actual results for various Vanguard bond funds.

In this first post is a chart of initial SEC yields and subsequent 5yr and 10yr returns for Vanguard California Intermediate-Term Tax-Exempt bond fund (one of the funds I own). I used investor shares (VCAIX) because there is more history for it than for admiral shares. Returns for 2005-2014 are from Vanguard, returns prior to that are based on adjusted-close prices from Yahoo Finance, which are not perfect but are pretty close (most returns I checked against Vanguard for 2005-2014 are within a couple of basis points).

Image

Anyone want to make any observations?

Kevin
The SEC Yield assumes that current holdings are held to maturity. An intermediate-term bond fund is unlikely to hold its securities until they mature. As their average maturity, and YTM, decrease they are sold and replaced with longer-term securities with higher YTMs. The SEC Yield is only useful in comparing one bond fund to another. Beyond that it has little predictive value as to the future yield of any particular bond fund.

billyt
Posts: 688
Joined: Wed Aug 06, 2008 9:57 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by billyt » Sun Jan 18, 2015 7:44 am

dkturner: You are misinformed. SEC yield is a very accurate predictor of subsequent 10 year return.

User avatar
galeno
Posts: 1576
Joined: Fri Dec 21, 2007 12:06 pm

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by galeno » Sun Jan 18, 2015 8:16 am

Correct.
billyt wrote:dkturner: You are misinformed. SEC yield is a very accurate predictor of subsequent 10 year return.
AA = 40/55/5. Expected CAGR = 3.8%. GSD (5y) = 6.2%. USD inflation (10 y) = 1.8%. AWR = 4.0%. TER = 0.4%. Port Yield = 2.82%. Term = 33 yr. FI Duration = 6.0 yr. Portfolio survival probability = 95%.

dkturner
Posts: 1509
Joined: Sun Feb 25, 2007 7:58 pm

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by dkturner » Sun Jan 18, 2015 8:19 am

billyt wrote:dkturner: You are misinformed. SEC yield is a very accurate predictor of subsequent 10 year return.
I believe that "current" yield is generally considered to be an accurate predictor of future total returns. "SEC" yield often differs wildly from "current" yield. Just for the hell of it why don't you take a look at Vanguard's intermediate-term bond funds and see how much of a variance there is between current (distribution) yield and SEC yield. Why is that? Which one is the more accurate predictor of future returns for each fund? They can't both be "right".
Last edited by dkturner on Sun Jan 18, 2015 8:31 am, edited 1 time in total.

billyt
Posts: 688
Joined: Wed Aug 06, 2008 9:57 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by billyt » Sun Jan 18, 2015 8:25 am

Are we looking at the same graph? The SEC yield and 10 year returns are within 1%. It has been show in a number of studies that the initial SEC yield explains 90% of the subsequent returns.

kenner
Posts: 3129
Joined: Sat Mar 01, 2008 8:45 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by kenner » Sun Jan 18, 2015 8:33 am

billyt wrote:Are we looking at the same graph? The SEC yield and 10 year returns are within 1%. It has been show in a number of studies that the initial SEC yield explains 90% of the subsequent returns.
It would be great if you could post links to some of those studies.

billyt
Posts: 688
Joined: Wed Aug 06, 2008 9:57 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by billyt » Sun Jan 18, 2015 8:42 am

Google is your friend. Start here: How to Predict the Next Decade's Bond Returns - WSJ. It is really a no-brainer. Virtually all bond returns come from the interest. There is a reason they call it fixed income.

billyt
Posts: 688
Joined: Wed Aug 06, 2008 9:57 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by billyt » Sun Jan 18, 2015 8:45 am

dkturner: Distribution yield can be higher or lower than SEC yield. Over time the tend to average out to the same thing.

lazyday
Posts: 3479
Joined: Wed Mar 14, 2007 10:27 pm

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by lazyday » Sun Jan 18, 2015 9:41 am

Don't know if this means anything:

A quick google says the maturity and duration of the fund is about 5 years, at least recently. But the 10 year returns were closer to sec yield than 5 year.

kenner
Posts: 3129
Joined: Sat Mar 01, 2008 8:45 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by kenner » Sun Jan 18, 2015 9:50 am

billyt wrote:Google is your friend. Start here: How to Predict the Next Decade's Bond Returns - WSJ. It is really a no-brainer. Virtually all bond returns come from the interest. There is a reason they call it fixed income.
Thank you , BillyT. Very helpful.

User avatar
Doc
Posts: 9226
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Doc » Sun Jan 18, 2015 9:55 am

billyt wrote:dkturner: You are misinformed. SEC yield is a very accurate predictor of subsequent 10 year return.
No. For the SEC yield to be a good predictor of subsequent returns one has to assume that the yield curve is static. The yield curve is of course not static. To paraphrase Larry Swedroe: "The current yield curve is the best predictor of the future yield curve. It's just not a very good one."

So while the SEC yield may be the best predictor of the future that we have it is still not a very good one. You also have the requirement of holding to maturity which many (most) funds do not do. This will make a significant difference unless the yield curve is flat which it is usually not.

As was pointed out up thread. The SEC yield is very good at comparing one fund to another but not so good at predicting future returns over longer periods.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

BigJohn
Posts: 1795
Joined: Wed Apr 02, 2014 11:27 pm

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by BigJohn » Sun Jan 18, 2015 11:02 am

billyt wrote:Google is your friend. Start here: How to Predict the Next Decade's Bond Returns - WSJ. It is really a no-brainer. Virtually all bond returns come from the interest. There is a reason they call it fixed income.
billyt, I read your suggested article but now I'm confused. In comments above you are arguing that SEC yield is the best predictor. However, the article is all about current yield as the best predictor. What am I missing??

User avatar
midareff
Posts: 6397
Joined: Mon Nov 29, 2010 10:43 am
Location: Biscayne Bay, South Florida

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by midareff » Sun Jan 18, 2015 11:06 am

billyt wrote:dkturner: You are misinformed. SEC yield is a very accurate predictor of subsequent 10 year return.

Don't you need a caveat here... such as in an environment where the interest rates DON'T CHANGE.

billyt
Posts: 688
Joined: Wed Aug 06, 2008 9:57 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by billyt » Sun Jan 18, 2015 11:49 am

No caveat needed. The historical record clearly shows that in spite of constantly changing interest rates and yield curves, the initial SEC yield is predicts the subsequent 10 year yield with a good degree of accuracy.

billyt
Posts: 688
Joined: Wed Aug 06, 2008 9:57 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by billyt » Sun Jan 18, 2015 11:51 am

And the SEC yield is simply a standardize way for expressing the current yield.

User avatar
midareff
Posts: 6397
Joined: Mon Nov 29, 2010 10:43 am
Location: Biscayne Bay, South Florida

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by midareff » Sun Jan 18, 2015 11:52 am

billyt wrote:No caveat needed. The historical record clearly shows that in spite of constantly changing interest rates and yield curves, the initial SEC yield is predicts the subsequent 10 year yield with a good degree of accuracy.

I just don't know what a backwards looking predictor tells me of value. Should I interpolate that bonds will produce zero for the next ten years, before taxes and CPI adjustments?

BigJohn
Posts: 1795
Joined: Wed Apr 02, 2014 11:27 pm

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by BigJohn » Sun Jan 18, 2015 12:00 pm

billyt wrote:And the SEC yield is simply a standardize way for expressing the current yield.
Now I'm really confused. Current yield is a dead easy calculation, why would it need standardization? My understanding is the SEC is a standardization to eliminate the need/desire to come to erroneous conclusions when comparing bonds funds based on current yield. So again, what am I missing??

richard
Posts: 7961
Joined: Tue Feb 20, 2007 3:38 pm
Contact:

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by richard » Sun Jan 18, 2015 12:07 pm

SEC yield is essentially yield to maturity (based on a 30 day average). The main alternative measure is current distribution yield.

YTM takes into account that current market price is different than par and that at maturity you get par. Current yield is just the amount distributed divided by price, without taking into account premium or discount to par.

User avatar
Doc
Posts: 9226
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Doc » Sun Jan 18, 2015 12:25 pm

BigJohn wrote:
billyt wrote:And the SEC yield is simply a standardize way for expressing the current yield.
Now I'm really confused. Current yield is a dead easy calculation, why would it need standardization? My understanding is the SEC is a standardization to eliminate the need/desire to come to erroneous conclusions when comparing bonds funds based on current yield. So again, what am I missing??
Because fund managers could manipulate the numbers by selling securities with premiums at the end of the accounting period to juice up their returns and negate their easily calculated but misleading current yield as a reasonable metric for comparison. The SEC yield is a mandate to eliminate or at least reduce that practice.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

User avatar
Doc
Posts: 9226
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Doc » Sun Jan 18, 2015 12:40 pm

midareff wrote:
billyt wrote:No caveat needed. The historical record clearly shows that in spite of constantly changing interest rates and yield curves, the initial SEC yield is predicts the subsequent 10 year yield with a good degree of accuracy.
I just don't know what a backwards looking predictor tells me of value. Should I interpolate that bonds will produce zero for the next ten years, before taxes and CPI adjustments?
Yes as long as the yield curve is static for the next ten years you will earn approximately zero. Oh gee, we all expect interest rates to go up we just don't know when so we don't think the curve will be static. = Caveat

Somewhere deep in the recesses I recall that the SEC yield was a good future predictor of returns about 30% of the time. The purpose of the SEC yield is to allow us poor investors to compare funds with some degree of reliability. It is not meant to be a predictor of the future.
Kevin M wrote:Anyone want to make any observations?
Perhaps the problem with Kevin's thread if there is one is that a while a CD is indeed a fixed income security a bond fund is not fixed. It is only the current portfolio that is fixed. It will change in the future. Bonds will mature and/or be sold and the new bods will likely have a different yield than those the fund has today. But Kevin started with Treasury bonds not funds. Somebody lead us astray somewhere and we lost the original path. :D
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

billyt
Posts: 688
Joined: Wed Aug 06, 2008 9:57 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by billyt » Sun Jan 18, 2015 12:47 pm

The SEC yield is forward looking. Bonds have a fixed face value and a fixed return. It doesn't take a rocket scientist to figure out their value or forward return. Bond desks trade these things furiously in an attempt to gain an advantage based on predictions of future interest rates. None of this trading affects the cash flow of the existing bonds. Yes, if you but a bond yielding 2% today, you might kick yourself if rates rise to 4% tomorrow (or congratulate yourself if rates fall to 1% tomorrow). However, none of this affects the fact that you will get your 2% and the face value of the bond back. A bond fund is a collection of bonds, and its behavior cannot stray very far from the behavior of the underlying investments. If you are a long term investor in a bond fund, dollar cost averaging in and out over your investing lifetime, you will earn the average return of the market over that time frame. If you are trading in and out of bond funds based on interest rate predictions, your return is less certain. The best estimate of the forward return at any point in time is the SEC yield.

BigJohn
Posts: 1795
Joined: Wed Apr 02, 2014 11:27 pm

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by BigJohn » Sun Jan 18, 2015 1:02 pm

Doc wrote:Because fund managers could manipulate the numbers by selling securities with premiums at the end of the accounting period to juice up their returns and negate their easily calculated but misleading current yield as a reasonable metric for comparison.
Thanks Doc, this was my understanding, just confused by the comment that SEC yield is a standard way to look at current . Current yield is what it is, SEC yield eliminates a bunch of potentially hidden variables to get to better comparison.

User avatar
Doc
Posts: 9226
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Doc » Sun Jan 18, 2015 2:01 pm

billyt wrote:The SEC yield is forward looking. Bonds have a fixed face value and a fixed return. It doesn't take a rocket scientist to figure out their value or forward return. Bond desks trade these things furiously in an attempt to gain an advantage based on predictions of future interest rates. None of this trading affects the cash flow of the existing bonds.
Right but those existing bonds disappear at some point in time. At some point the manager sells a bond or it matures. Then you have a new bond and the cash flow is different. It is this situation that limits the applicability of using the SEC yield as a forward looking metric. As long as that new bond looks like the old one everything is fine. But that usually isn't the case. Maybe the manager is making trades to try to improve his return. Or maybe he has to sell bonds because his charter is for a 5-10 bond maturity. If the new bond is different from the old one the SEC yield is no longer valid as a forward looking metric. In order to use the SEC yield as a forward predictor you have to assume that the manager keeps operating as he did in the past and that the yield curve remains static. The manager might be very consistent in his philosophy as with an index fund but he cannot make the yield curve never change. Bonds may invest in fixed income securities but the income from bond funds is not fixed.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

User avatar
ogd
Posts: 4875
Joined: Thu Jun 14, 2012 11:43 pm

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by ogd » Sun Jan 18, 2015 2:02 pm

Kevin: there were also the graphs produced by user Ranger in this older thread you were on: http://www.bogleheads.org/forum/viewtop ... 7#p1758761 , from a combination of the Agg bond index and Ibbotson's data, very long term back to the 30s.

Unfortunately, imageshack went south since then and I don't thing Ranger is around anymore. But my recollection of the nominal graph was "fits like a glove"; the inflation-adjusted graph was wilder but ended up looking pretty good too. I also found the Vanguard graph you posted there reassuring.

Unchanging interest rates are not an assumption here, in fact data going back to 70s or longer sees some pretty wild swings in rates. It's just how things work out with bonds, because they are so predictable. And yes, that includes funds. Worth remembering, though, that a 10 year return period is about twice as long as the duration of the funds and indices in question, which makes things considerably smoother.

User avatar
Topic Author
Kevin M
Posts: 11183
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Kevin M » Sun Jan 18, 2015 2:44 pm

kenner wrote: It would be great if you could post links to some of those studies.
I've done my own!

Bogleheads • View topic - Initial yield and subsequent N-year return. Granted, this is for a 5-year constant maturity Treasury, but we can very roughly think of that as similar to a bond fund with constant 5-year maturity. Note that I found correlation of 0.85 both for subsequent 5-year and 10-year returns, which is pretty close to the 0.90 (or 0.92 or whatever) result for US aggregate bond market.

Note also however that high correlation doesn't mean no surprises; with sustained, multi-year increase or decrease in the relevant rate(s), the subsequent 5-year or 10-year returns can deviate significantly from initial yield to maturity. We see that even with the limited data in the OP.

Note that the larger deviations are not "1%", but on the order one percentage point of annualized return. Although it's common usage to say one percent when we mean one percentage point (and even I do so when I think it's clear in context), in cases like this it's worth distinguishing between them.

If initial yield is 3% and we earn a 2% annualized return, that's one percentage point difference, but the subsequent n-year annualized return is 67% (2/3) of what we expected, so 33% lower. Or to look at the positive "surprises", if initial yield is 3% and our subsequent n-year return is 4%, we earned one percentage point more, but 33% more than "expected".

Since the annualized difference compounds, over five years a 3% annualized return gives us 5% 53% more cumulative return than a 2% annualized return, and over 10 years it gives us 10.25% 57% more*. Not something to sneeze at.

*EDIT: 2% compounded for 5 years is 10.4%, and 3% for 5 years is 15.9%; 15.9/10.4 = 1.53 = 53% more. 2% compounded for 10 years is 21.9%, and 3% compounded for 10 years is 34.4%; 34.4/21.9 = 1.57 = 57% more.

Kevin
Last edited by Kevin M on Mon Jan 19, 2015 9:35 pm, edited 1 time in total.
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Topic Author
Kevin M
Posts: 11183
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Kevin M » Sun Jan 18, 2015 3:31 pm

With respect to the discussions about the differences between SEC yield, yield to maturity (YTM), average YTM, current yield, and distribution yield, I started this thread to try and clarify and provide a thread to discuss exactly these things: Bogleheads • View topic - SEC Yield, YTM, Distribution Yield, Current Yield.

To summarize ...

For an individual bond, "current yield" generally refers to the sum of annual coupon payments divided by the current value of the bond. The analog to this for a bond fund is distribution yield, but there are different ways to calculate it, and it generally includes accrual/accretion of discounts/premiums.

The way Vanguard calculates distribution yield comes pretty close to the way you'd calculate current yield for a bond. The latest distribution amount is divided by the average fund price over previous 30 days, then annualized (divide by days in month, multiply by 365). By contrast, Morningstar publishes the trailing 12-month (TTM) distribution yield. Since there is no fixed coupon payment for a bond fund, and since discounts and premiums are factored in, there really isn't an exact bond fund analog to current yield for a bond.

For an individual bond, yield to maturity (YTM)( is the single discount rate that equalizes discounted cash flows and current price. Vanguard reports average YTM for its bond funds, which is a weighted average of the YTMs of the fund's bond holdings, and for some funds factors in the possibility of bonds being called. For an individual bond, yield-to-worst factors in the bond being called, but YTM does not.

SEC yield is a standardized method of calculating average YTM, averaged over the previous 30 days, and subtracting fund expenses. You can look at SEC yield and average YTM for Vanguard's bond funds; they are likely to be pretty close for most bond funds if you look at them on the same date, but there definitely can be some variation. SEC yield is updated daily, but average YTM is only reported as of certain dates (e.g., you can see it now for 12/31/2014). As far as I know, Vanguard only provides historical data for SEC Yield, so that's what I use.

Vanguard also reports average coupon rate for its bond funds. If the average coupon rate is higher than the average YTM, the distribution yield probably is higher than the SEC yield, but there won't be a precise relationship because of all the averaging and different ways these various numbers are calculated.

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

billyt
Posts: 688
Joined: Wed Aug 06, 2008 9:57 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by billyt » Sun Jan 18, 2015 6:00 pm

Kevin: To call the difference between 2% and 3% a 33% difference, while technically correct, is a ridiculous way to look at it. Over the 10 years, your returns will be similar to the initial SEC yield, that is the bottom line. Moreover, although there may be a difference between the initial yield and the subsequent ten year return for any particular 10 year interval, these differences will become smaller and smaller if you start summing up the rolling 10 year intervals. The bottom line is that over long periods of time, nearly 100% of the return of a bond fund comes from the interest. Active managers struggle to add a few basis points per year of extra return by anticipating interest rate moves and often fail. An index fund will return you the average of the market rates over your investing lifetime.

User avatar
Doc
Posts: 9226
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Doc » Sun Jan 18, 2015 6:17 pm

billyt wrote: Over the 10 years, your returns will be similar to the initial SEC yield, that is the bottom line.
The current SEC Yield for Vanguard Total Bond Market 1.87% (Morningstar). Do you really believe that the average return on TBM will be close to 1.87% over the next ten years?
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

billyt
Posts: 688
Joined: Wed Aug 06, 2008 9:57 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by billyt » Sun Jan 18, 2015 6:28 pm

Yes it will. If rates go up, the NAV will fall to and you will have to wait to earn that 1.87% return. If rates fall, you will get an early windfall, but earn less going forward. Shares of total bond that you buy today will earn close to 1.87% over the next ten years. Hopefully, rates will increase and you can buy more shares that will earn a higher return. One sure way to increase your return is to buy Admiral shares (earning 2%).

User avatar
Topic Author
Kevin M
Posts: 11183
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Kevin M » Sun Jan 18, 2015 6:31 pm

billyt wrote:Kevin: To call the difference between 2% and 3% a 33% difference, while technically correct, is a ridiculous way to look at it.
Perhaps, but it's just as Not at all. What's ridiculous is to characterize it as a 1% difference, when the cumulative impact over 10 years is a difference of more than 10% 57%! Looking at the cumulative impact probably is the most sensible way to evaluate it. Small differences in annualized returns can make big differences in cumulative returns.
billyt wrote:Over the 10 years, your returns will be similar to the initial SEC yield, that is the bottom line.
One of the reasons I started this thread (and others like it) is so that we can evaluate these generalized statements by looking at actual data. The data shows that this statement is false, unless you believe that $110,000 is "similar to" $100,000, and/or you believe that "will be" is the same as "is likely to be".

We might as well say that an index fund with an expense ratio of 1.1% will have returns similar to a similar fund with an ER of 0.1%, so don't worry so much about that tiny "1%" difference in ER. I'm not saying that these are directly analogous, but using it to illustrate that "similar" is not "equal to", and numbers that appear small when framed one way appear larger when framed another way.

I don't see that any of your other comments are addressing the topic of this thread, which is simply to evaluate how predictive SEC yield is of subsequent 5-year and 10-year returns. I'm very familiar with your views on bond funds, and there are plenty of threads where you can continue to propagate your beliefs. The purpose of this thread is to see what the data shows us, pure and simple.

If you have some additional data that shows the relationship of initial SEC yield to subsequent returns, I'd love to see it. I plan to post more myself.

Kevin
Last edited by Kevin M on Mon Jan 19, 2015 9:38 pm, edited 1 time in total.
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

billyt
Posts: 688
Joined: Wed Aug 06, 2008 9:57 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by billyt » Sun Jan 18, 2015 6:49 pm

The idea that nearly 100% of bond fund returns come from the interest is not mine, but comes from John Bogles "Little Book of Common Sense Investing." It is indeed common sense. If this is true, then over time, the average return must converge with the average SEC yield. I don't know how anyone could imagine it would be otherwise. The data support this conclusion. Where is that graph from Vanguard that you posted a while ago that shows this very clearly?

billyt
Posts: 688
Joined: Wed Aug 06, 2008 9:57 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by billyt » Sun Jan 18, 2015 7:05 pm

Furthermore, the reason that we focus on small differences in expense ratios is because this is something that we can control. It make sense to go for those extra basis points that Admiral shares afford. Market returns on the other hand are completely beyond any investors control. As far as predicting future returns of a fund is concerned, if you can guess the annualized future return within a percent or two, that is pretty good and perfectly adequate for planning purposes. Contrast this with trying to predict the annualized return of a stock fund over 10 years!

stlutz
Posts: 5422
Joined: Fri Jan 02, 2009 1:08 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by stlutz » Sun Jan 18, 2015 7:27 pm

Instead of using real world returns I decided to try a few hypothetical scenarios. My simplifying assumption was to buy a 6 year bond, hold for one year, sell, rinse and repeat for 10 years. This gives my portfolio an average duration that approximates many IT Treasury funds (i.e. a little over 5).

Scenario 1: starting rate of 2% with a .2% "slope" (i.e. 5 year bond yielding 1.8% at the start of the exercise). Rates increase .5% every year for 10 years, with the slope staying the same. Starting SEC Yield = 2%; 10 year rate of return = 2.9%/yr.

Scenario 2: Same as scenario 1 but rates go up 1.5% each and every year: SEC Yield = 2%; 10yr return = 3.58%

Scenario 3: Same as scenario 1 but rates stay unchanged the entire time: SEC Yield=2%; 10 yr return = 2.95%

Scenario 4: The reverse of 1. Rates start at 8%. The curve is inverted so the 5 year rate at the start is 8.2%. Rates drop by .5% every year; the curve stays inverted. SEC Yield = 8%; 10 year return = 4.46%

Scenario 5: Same as #4 except rates stay unchanged the entire time: SEC Yield = 8%; 10 yr. return = 7.19%

Scenario 6: "Bondmageddon". We start at the same place as scenario 2. But, in the second year rates go up to 7% and the yield curve becomes completely flat. Rates stay unchanged from there forward. SEC Yield = 2%; 10 Year return = 3.78%.

Conclusions:

These are all artificial scenarios, but the illustrate a few things:

a) Initial SEC Yield is a number the the returns revolve around, but the error margin is pretty large.
b) If your holding period is double the duration of the fund, you are much better off if rates go up--the more the better.
c) The shape of the yield curve is an important component of return.

In the real world where things aren't as linear as my scenarios above, and the actual results will be closer to the initial yield. It's in the eye of the beholder, however, as to whether something like a 1%/yr. error margin makes for an "accurate" projection or not.

User avatar
Topic Author
Kevin M
Posts: 11183
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Kevin M » Sun Jan 18, 2015 8:09 pm

billyt wrote:The idea that nearly 100% of bond fund returns come from the interest is not mine, but comes from John Bogles "Little Book of Common Sense Investing."
You seem to want to continue to emphasize a particular point of view rather than discuss the topic of the thread. Your statement may be true, but that's not the discussion here.

JB also shows a matrix in one of his books that demonstrates how much return comes from interest and now much from price change (I think he uses a 20-year bond held for 10 years). So he doesn't just completely ignore the latter. This thread is more in the vein of that chart rather than the more generalized statements. Here the idea is to dig into the details.
billyt wrote:Where is that graph from Vanguard that you posted a while ago that shows this very clearly?
I think the graph you're talking about is almost exactly like the one in the OP, and like ones I've posted in other threads and for longer time periods. I'll be sharing more of them here. I've also posted the correlation charts, which is another good way to look at it, but I wanted to start simple here.

You seem to be missing the point that there's no disagreement (from me at least) that there is high correlation between initial yield to maturity and subsequent 10-year return. As I recall the correlation is 92% for the aggregate US bond market. I got 85% when looking at 5-year constant maturity Treasuries since the 1950s, so in the same ballpark.

One important point that is relevant in this thread is that high correlation is not a guarantee, and that you can have high correlation with significant deviations. I think some people will be interested in seeing how this has worked out historically, so they can be comfortable that their bond fund 10-year return going forward is likely to be in the ballpark of current SEC yield, let's say 2% for aggregate US bond, but it is by no means guaranteed.

As a matter of fact, for Vanguard Total Bond Index fund I found the average subsequent 5-year return premium over SEC yield to be about 50 basis points. But we also have to remember that this is for a period when rates have generally declined. I was planning on re-posting the relevant chart or charts in this thread at some point.

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

User avatar
Topic Author
Kevin M
Posts: 11183
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Kevin M » Sun Jan 18, 2015 8:25 pm

stlutz wrote:Instead of using real world returns I decided to try a few hypothetical scenarios. My simplifying assumption was to buy a 6 year bond, hold for one year, sell, rinse and repeat for 10 years.
This seems very similar in concept to evaluating yield and subsequent returns for the 5-year constant-maturity Treasury (rolled over annually), except that you're using a 6-year bond. I've already referenced the thread in which I shared the results for all history available (Bogleheads • View topic - Initial yield and subsequent N-year return), but I might as well at least show a couple of the charts again here.

Here is the initial yield and subsequent 5-year returns chart:

Image

And here's the chart for subsequent 10-year returns:

Image

As already noted, although the 10-year returns chart looks like a better fit, the correlation actually is about the same as for the 5-year returns: 0.85%.

Kevin
Last edited by Kevin M on Mon Jul 23, 2018 12:52 pm, edited 1 time in total.
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

mindbogle
Posts: 140
Joined: Sun Feb 10, 2013 11:28 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by mindbogle » Sun Jan 18, 2015 10:58 pm

Kevin M wrote:
stlutz wrote:Instead of using real world returns I decided to try a few hypothetical scenarios. My simplifying assumption was to buy a 6 year bond, hold for one year, sell, rinse and repeat for 10 years.
This seems very similar in concept to evaluating yield and subsequent returns for the 5-year constant-maturity Treasury (rolled over annually), except that you're using a 6-year bond. I've already referenced the thread in which I shared the results for all history available (Bogleheads • View topic - Initial yield and subsequent N-year return), but I might as well at least show a couple of the charts again here.
Kevin, thanks for starting another interesting bond thread.

Does Damodaran implicitly assume a flat yield curve in his total return calculations, or does he take into account the historical shape of the yield curve? I see stlutz is rolling down a hypothetical yield curve in his calculations. If the Damodaran total return calculations do not take yield curve shape into account, then you may be overstating the degree to which YTM foreshadows future total return. Can you provide more detail on how returns are calculated for Damodaran's data?

Another source for understanding the linkage between YTM and future returns is Ryan Lab's Treasury Indexes found here (registration required). There you will find (almost) constant maturity treasury index data constructed by rolling over actual bonds to maintain maturity, starting in the 1970's (no more need for the "stlutz" fund data!). The index total returns are derived from bond prices and thus historical yield curve shape is accounted for. Spreadsheets include all the needed index fundamentals (total return, YTM, duration, maturity, coupon) to easily compare YTM to future returns. When I explored this data a couple of years back, I concluded that YTM was a slightly better predictor of future maturity-yr total returns than it was for future duration-yr returns. Also, I recall the prediction errors seemed to be greater magnitude (and correlations lower) than what you posted using Damodaran's data (the reason for my inquiry above).

Thanks,

MB

User avatar
Topic Author
Kevin M
Posts: 11183
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Kevin M » Mon Jan 19, 2015 12:08 am

mindbogle wrote: Does Damodaran implicitly assume a flat yield curve in his total return calculations, or does he take into account the historical shape of the yield curve?
Good question. I think the answer is flat yield curve, since it's the 5-year rates that are used in the price change part of the return calculation. So maybe a better analogy would be rolling a 5-year bond daily with no transaction costs.

I don't want to get too deep into that in this thread, since the focus here is on bond fund SEC yield and returns. We can go deeper into the CMT data and calculations in the linked thread if you want.
mindbogle wrote:There you will find (almost) constant maturity treasury index data constructed by rolling over actual bonds to maintain maturity, starting in the 1970's.
The thing is you have to go back to the 1950s, and preferably 1940s to see an environment in which we had a prolonged period of low rates, comparable to today's low rates, followed by rising rates at some point. I think it's more educational to look at a period like that to get an idea of what could happen going forward from here. What we know can't happen is anything that's happened (in the US anyway) since the 1970s, and certainly not since the 1980s. Yet that is the period for which we can get bond fund data, and that I think tends to influence people's views on bonds (recency bias).

I guess we could use 5-year and 3-year CMT data (and 10-year if we want) back to 1953 to better simulate a rolling bond ladder, but again, that's probably more a topic for the other thread.
mindbogle wrote:Also, I recall the prediction errors seemed to be greater magnitude (and correlations lower) than what you posted using Damodaran's data (the reason for my inquiry above).
But the interesting thing is that the correlation is even higher for US aggregate bond market since the early 1970s. IIRC, it's about 92%. The purpose of this thread is to see how well various Vanguard bond funds fit this model--at least for the relatively short time periods we have data for.

Not that we have enough data for it to be very meaningful, but since 1995, correlation for the CA muni bond fund featured in the OP is 0.77 for 5-year returns and 0.83 for 10-year returns. For Vanguard Total Bond Index fund since 1993 correlation is 0.81 for 5-year returns and 0.83 for 10-year returns. So all these numbers are roughly in the same ballpark of 0.8.

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

longinvest
Posts: 3895
Joined: Sat Aug 11, 2012 8:44 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by longinvest » Mon Jan 19, 2015 8:35 am

Related thread that explains where bond fund returns come from:

How can Bond Funds perform better than yield
Bogleheads investment philosophy | single-ETF balanced portfolio (VBAL) | VPW accumulation

mindbogle
Posts: 140
Joined: Sun Feb 10, 2013 11:28 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by mindbogle » Mon Jan 19, 2015 8:59 am

Kevin M wrote:
mindbogle wrote: Does Damodaran implicitly assume a flat yield curve in his total return calculations, or does he take into account the historical shape of the yield curve?
Good question. I think the answer is flat yield curve, since it's the 5-year rates that are used in the price change part of the return calculation.
I think it is important if we are trying to assess the degree to which YTM (or SEC yield) explains future bond returns. Despite the back and forth debate in this thread, I think all agree that YTM does explain a large component of future returns - seem to be arguing degree.
Kevin M wrote: So maybe a better analogy would be rolling a 5-year bond daily with no transaction costs.
An index constructed by rolling bonds daily (versus monthly or annually) does not reduce the impact of yield curve shape on returns. It wouldn't be all that hard to incorporate yield curve shape into a hypothetical constant maturity index going back as far as the yield data exists. Maybe in my spare time....

MB

User avatar
Doc
Posts: 9226
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Doc » Mon Jan 19, 2015 9:34 am

mindbogle wrote:I think it is important if we are trying to assess the degree to which YTM (or SEC yield) explains future bond returns. Despite the back and forth debate in this thread, I think all agree that YTM does explain a large component of future returns - seem to be arguing degree.
Thinking out loud.

Duration is often thought about as the price sensitivity to changes in market rates but it is also the "time to indifference" and the time "it takes to get your money back". With this in mind I would think that the SEC yield will explain the future bond fund return for the duration. Indeed if the fund never replaced maturing bonds it would be exact. And if the the fund replaced bonds with similar issues both in duration and YTM it would also be exact. In fact if the yield curve is static the fund would look like a rolling ladder and the SEC yield and bond return would go on together forever. So for an index fund that isn't changing its portfolio much the only difference in SEC yield and future bond fund return is due to shifts in the yield curve. So what Kevin's chart is showing is mostly the effect of a changing yield curve. :idea: :?:
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

longinvest
Posts: 3895
Joined: Sat Aug 11, 2012 8:44 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by longinvest » Mon Jan 19, 2015 9:35 am

mindbogle wrote:
Kevin M wrote:
mindbogle wrote: Does Damodaran implicitly assume a flat yield curve in his total return calculations, or does he take into account the historical shape of the yield curve?
Good question. I think the answer is flat yield curve, since it's the 5-year rates that are used in the price change part of the return calculation.
I think it is important if we are trying to assess the degree to which YTM (or SEC yield) explains future bond returns. Despite the back and forth debate in this thread, I think all agree that YTM does explain a large component of future returns - seem to be arguing degree.
Kevin M wrote: So maybe a better analogy would be rolling a 5-year bond daily with no transaction costs.
An index constructed by rolling bonds daily (versus monthly or annually) does not reduce the impact of yield curve shape on returns. It wouldn't be all that hard to incorporate yield curve shape into a hypothetical constant maturity index going back as far as the yield data exists. Maybe in my spare time....

MB
MB, Maybe you'll like the thread linked-to in my post ( viewtopic.php?uid=39265&f=10&t=155923&start=0#p2340120 ).
Bogleheads investment philosophy | single-ETF balanced portfolio (VBAL) | VPW accumulation

User avatar
Topic Author
Kevin M
Posts: 11183
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Kevin M » Mon Jan 19, 2015 1:47 pm

mindbogle wrote: An index constructed by rolling bonds daily (versus monthly or annually) does not reduce the impact of yield curve shape on returns.
Yes, but then doesn't the shape only matter between the 2 days for each bond? If we assume the curve is linear between the points for which we can get long-term data (e.g., 3-year, 5-year, 10-year), then I agree that the rolling interval doesn't matter, as long as it's within one of the linear segments (e.g., between 3 and 5 or 5 and 10 years).
mindbogle wrote:It wouldn't be all that hard to incorporate yield curve shape into a hypothetical constant maturity index going back as far as the yield data exists. Maybe in my spare time....
If you tell me what your are thinking, maybe I can help. I was thinking of assuming linear curve between the terms for which we can get historical CMT data back to 1953 (7-year only goes back to 1969). Then price change would be calculated using the say the 5-year rate for Year N and the linearly interpolated 4-year rate for Year N+1, using the 3-year and 5-year CMT data to interpolate the 4-year rate. Or use the 5-year and 10-year rates to interpolate rates for terms of 6, 7, 8 or 9 years.
mindbogle wrote:MB
Where were you when I posted this:
Initial yield and subsequent N-year return??? This is where I wanted to have these discussions.

My concern is that if we get too deep and technical here, we lose more people who might otherwise stay tuned in if we look at actual Vanguard funds that they may actually own.

Any votes? Go as deep as we want here? Or move the more technical investigations to the other thread that was intended to serve that purpose, and here stay focused on SEC yield and Bond fund returns? Lurkers please vote too (or would that be a paradox?).

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

User avatar
Topic Author
Kevin M
Posts: 11183
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Kevin M » Mon Jan 19, 2015 2:10 pm

Doc wrote: Duration is often thought about as the price sensitivity to changes in market rates but it is also the "time to indifference" and the time "it takes to get your money back".
This only works for an individual bond. Since a typical bond fund does not mature, as does an individual bond, you have to reset your indifference clock for each rate change.
Doc wrote:With this in mind I would think that the SEC yield will explain the future bond fund return for the duration.
Maybe it will work better, which maybe is why the studies like Vanguard's look at 10-year returns instead of 5-year returns. But the evidence I've looked at so far, all of which has been shared here or in other threads, indicates that whether you look at 5 years or 10 years, you get similar results in terms of correlation. But at least visually, the 10-year return charts seem to show what appears to be better tracking.
Doc wrote:So for an index fund that isn't changing its portfolio much the only difference in SEC yield and future bond fund return is due to shifts in the yield curve. So what Kevin's chart is showing is mostly the effect of a changing yield curve. :idea: :?:
Yes and no. In the thread that longinvest has linked to a couple of times, I share my personal discovery (that others already had discovered) that average YTM understates the subsequent 1-year return of a rolling bond ladder, assuming static yield curve: How can Bond Funds perform better than yield (average YTM understates the subsequent 1-year return of a rolling bond ladder).

Of course changes in yield curve have an effect too. One of the observations I was waiting to hear someone make is that for the fund featured in the OP, the only "big" 5-year return surprise on the downside was for the 5-year period starting in 2004--a period when rates moved generally higher (up a little, up "a lot", flat, flat, up "a lot", where "a lot" is 40 basis points or more).

Conversely, during all other 5-year periods, and all 10-year periods rates generally declined, so perhaps it's not surprising that returns were generally higher during the period for which we have fund data. This is consistent with the general trends we see if looking at the 5-year or 10-year CMT data back to 1953 (without factoring in yield curve shape, as MB has highlighted).

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

mindbogle
Posts: 140
Joined: Sun Feb 10, 2013 11:28 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by mindbogle » Mon Jan 19, 2015 3:41 pm

Kevin M wrote:
mindbogle wrote: An index constructed by rolling bonds daily (versus monthly or annually) does not reduce the impact of yield curve shape on returns.
Yes, but then doesn't the shape only matter between the 2 days for each bond? If we assume the curve is linear between the points for which we can get long-term data (e.g., 3-year, 5-year, 10-year), then I agree that the rolling interval doesn't matter, as long as it's within one of the linear segments (e.g., between 3 and 5 or 5 and 10 years).


I didn't say the rolling interval "doesn't matter". I said that making the interval shorter doesn't reduce the impact of yield curve shape. I may have misunderstood your "analogy" earlier - it sounds like we are both on the same page now.
Kevin M wrote:
mindbogle wrote:It wouldn't be all that hard to incorporate yield curve shape into a hypothetical constant maturity index going back as far as the yield data exists. Maybe in my spare time....
If you tell me what your are thinking, maybe I can help. I was thinking of assuming linear curve between the terms for which we can get historical CMT data back to 1953 (7-year only goes back to 1969). Then price change would be calculated using the say the 5-year rate for Year N and the linearly interpolated 4-year rate for Year N+1, using the 3-year and 5-year CMT data to interpolate the 4-year rate. Or use the 5-year and 10-year rates to interpolate rates for terms of 6, 7, 8 or 9 years.

Yes, assuming linear segments sounds reasonable to me and should get us a lot closer to what a semi-constant maturity fund total return would look like.
Kevin M wrote:
mindbogle wrote:MB
Where were you when I posted this:
Initial yield and subsequent N-year return??? This is where I wanted to have these discussions.

My concern is that if we get too deep and technical here, we lose more people who might otherwise stay tuned in if we look at actual Vanguard funds that they may actually own.

Yeah, I missed that thread during my recent BH sabbatical (but you are the one who linked to it! :happy). If I come up with anything more, I'll post over there so you can stay on your intended point here.

MB

stlutz
Posts: 5422
Joined: Fri Jan 02, 2009 1:08 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by stlutz » Mon Jan 19, 2015 4:10 pm

One of the observations I was waiting to hear someone make is that for the fund featured in the OP, the only "big" 5-year return surprise on the downside was for the 5-year period starting in 2004--a period when rates moved generally higher (up a little, up "a lot", flat, flat, up "a lot", where "a lot" is 40 basis points or more).

Conversely, during all other 5-year periods, and all 10-year periods rates generally declined, so perhaps it's not surprising that returns were generally higher during the period for which we have fund data. This is consistent with the general trends we see if looking at the 5-year or 10-year CMT data back to 1953 (without factoring in yield curve shape, as MB has highlighted).
Comparing that to the fake 10-year scenarios I constructed above does reveal one interesting point. If your holding period is only the duration of the fund, then it's bad for you if rates steadily go up. On the other hand, if your holding period is double the duration of the fund, it seems you're better off if rates go up.

As such, it would seem that for the person who is worried about the impact of rising rates but still wants to get at least the SEC Yield return, their target holding period should be duration x 2, not duration x 1. (Although I guess that wouldn't hold if my "bondmageddon" scenario had rates going up in year 9 instead of year 2).

User avatar
Electron
Posts: 1908
Joined: Sat Mar 10, 2007 8:46 pm

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by Electron » Mon Jan 19, 2015 7:12 pm

Kevin M wrote:The thing is you have to go back to the 1950s, and preferably 1940s to see an environment in which we had a prolonged period of low rates, comparable to today's low rates, followed by rising rates at some point. I think it's more educational to look at a period like that to get an idea of what could happen going forward from here. What we know can't happen is anything that's happened (in the US anyway) since the 1970s, and certainly not since the 1980s. Yet that is the period for which we can get bond fund data, and that I think tends to influence people's views on bonds (recency bias).
I had the same exact thoughts. It would be very interesting to see the subsequent returns on a similar chart in a period of rising rates. The periods of falling rates are also very interesting to study.

Your charts could still offer a clue. There have been brief periods of rising rates fairly recently. One example would be shorter term rates in the period 2004 through 2007.

http://research.stlouisfed.org/fred2/series/TB3MS

It's not clear how much that period affected municipal bond intermediate term rates. If there was an effect, it could show up as a temporary deviation in the 5 and 10 year forward returns on the chart. Perhaps that rise in rates would show up in the forward returns of short or limited maturity funds.
Electron

haginile
Posts: 31
Joined: Tue Aug 12, 2014 3:03 pm
Contact:

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by haginile » Mon Jan 19, 2015 8:58 pm

I did some research last year on this, and the conclusion is largely consistent with the discussion above: starting yield is a remarkably good forecasts for subsequent returns.

I think the prevalent belief (one I used to held too...) is that if bonds are not held to maturity, duration impact could cause return to deviate from yield. If reality: when yields go up, you lose money because of duration; but then you can reinvest at higher yields. Conversely, when yields go down, you have capital gains, but you must now reinvest cash flows at lower yields. Over a long horizon, these effects tend to wash out.

The chart below shows the starting 10-year par yield and the subsequently realized 10-year return of 10-year par bonds (rolled monthly). I also show the rolling yield (starting yield + roll down return), which proves to be an even better predictor.

Image

User avatar
Topic Author
Kevin M
Posts: 11183
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Initial SEC yield and subsequent 10yr returns:VBMFX

Post by Kevin M » Mon Jan 19, 2015 9:04 pm

OK, trying to get this thread back on track, here are results for initial SEC yield and subsequent 10-year return for Vanguard Total Bond Market Index Fund (VBMFX):

Image

So again, we see rough tracking, but significant deviations. The correlation of subsequent 10-year return to initial yield is 0.82, so not as high as the 0.92 number that we've seen for aggregate US bond market since 1972 (or whenever), but still pretty high.

I think it's also informative to view the subsequent 10-year return minus the initial SEC yield:

Image

From this it looks that it's more likely for subsequent annualized return to be greater than +/- 50 basis points relative to initial SEC yield than it is to be equal to initial SEC yield, and a deviation of 100 basis points shouldn't be particularly surprising. I think this is more informative than a single number like correlation of 0.82.

Back to the point of considering "only 1%" difference in annualized return as a small number, I actually under-emphasized the impact in responding to this point previously. An annualized return of 1% results in a cumulative 10-year return of 10.5%, while an annualized return of 2% results in a cumulative 10-year return of 21.9%, so the latter provides a cumulative return of more than 2X the former. So we better hope that the 10-year return of our intermediate-term bond funds is not "only 1%" below current SEC yield.

The positive outlook is that a 3% annualized return provides a cumulative 10-year return of 34.4%, so about 1.57X better than a 2% annualized return. So let's hope that rates rise quickly, we take a big hit soon, and end up much better off in 10 years. I'm sure many folks would freak out initially if rates increased a lot quickly (and we won't be seeing posts about how much better intermediate-term bond fund returns have been than CDs or short-term bond funds), but we really will be better off in the long run.

Kevin
Last edited by Kevin M on Mon Jul 23, 2018 12:57 pm, edited 2 times in total.
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

mindbogle
Posts: 140
Joined: Sun Feb 10, 2013 11:28 am

Re: Initial SEC yield and subsequent 5yr and 10yr returns

Post by mindbogle » Mon Jan 19, 2015 9:17 pm

haginile wrote:I did some research last year on this, and the conclusion is largely consistent with the discussion above: starting yield is a remarkably good forecasts for subsequent returns.

I think the prevalent belief (one I used to held too...) is that if bonds are not held to maturity, duration impact could cause return to deviate from yield. If reality: when yields go up, you lose money because of duration; but then you can reinvest at higher yields. Conversely, when yields go down, you have capital gains, but you must now reinvest cash flows at lower yields. Over a long horizon, these effects tend to wash out.

The chart below shows the starting 10-year par yield and the subsequently realized 10-year return of 10-year par bonds (rolled monthly). I also show the rolling yield (starting yield + roll down return), which proves to be an even better predictor.

Image
Excellent post (although Kevin may complain that you are hijacking the OP!). This may be exactly what Kevin and I were talking about earlier in this thread. How are you defining "roll down return" for your rolling yield? Please give more details so that we can evaluate the relevance of the plot shown. Can you also post cross-plots of "rolling return" and initial par yield against subsequent 10-yr return? Indeed, visually, the roll-return looks like it correlates significantly better to subsequent returns.

Thanks,

MB

Post Reply