Grok's Tip#9: To Manage Risk Well, Use a Bar-Bell

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Grok's Tip#9: To Manage Risk Well, Use a Bar-Bell

Post by grok87 » Thu Apr 28, 2011 3:23 pm

Grok's Investment Tip#9/10: To Manage Risk Well, Use a Bar-Bell

What is a Barbell? In fixed income investing, a “barbell” strategy is a combination of a short term bond (or cash) and a long term bond. For example say t-bills and a 10 year treasury. It is the opposite of the “bullet” strategy of just buying an intermediate term bond, for example the 5 year treasury. Due to the way bonds work, barbelling often provides more downside protection in the case of interest rate shocks. More generally, the concept of barbelling can be used to reduce or better manage risk for other investment strategies and even outside the investment world. Let’s work through some examples, starting with the traditional bond ones:


a) Bond Barbells

Let's say you want to stay relatively short in your fixed income duration. You could just buy the 3 year treasury. Here's what you would get:

Bond................Yield.....Duration.....Loss from a 4% interest rate shock
3 yr. treasury....1.02%......2.96.........11.0%

The last item means that if interest rates rise by 4 percentage points then your investment will lose 11% in market value. The 4 percentage point shock is what the fed is currently using to evaluate bank risk management (is that an Oxymoron?)

Alternatively you could "barbell" by purchasing a 46/54 combination of a 7 year treasury and cash. This would have the same duration of 2.96 as the 3 year treasury. But you get a better yield of 1.36% vs. 1.02% for the 3 year. Plus your market value hit is less in the case of the 4% point interest rate shock, only 10% vs. 11% for the 3 year treasury.

Item.................Yield.....Duration.....Loss from a 4% interest rate shock
7 year treasury...2.96%.....6.43..........22.1%
tbill....................0.00%.....0.00..........0.0%
46/54 combo.......1.36%.....2.96..........10.2%

This combination of better expected returns and lower risk is what is often referred to as a free lunch (who says there are none!)

Here's another example of a bond barbell that I like. Instead of buying the Vanguard Intermediate Tax Exempt bond fund with a yield of 3.21% and a duration of 5.6, buy an 80/20 mix of the Vanguard Long Term Tax Exempt bond fund (yield 4.02%, duration =7.0) and an FDIC insured savings account (pre-tax yield 1%, after-tax yield 0.65%). The 80/20 mix has the same duration but a better yield of 3.35%. Plus the credit quality is better as 20% is in US government guaranteed investments.


b) Asset Allocation Barbelling.

Larry Swedroe has been talking about this concept for a while now. Here's my example: Instead of a traditional 60/40 mix of the S&P 500 and the Barclays Agg. Bond index, buy a 40/60 mix of Small Cap Value stocks and intermediate treasuries. Why is this barbelling? Well the S&P 500 is a moderately risky asset as far as equities go and so is the Barclays Agg. in the fixed income space. Small Cap Value stocks are generally viewed as more risky equities (the market hates those companies, hence their low market cap and valuation) whereas Treasuries are super-safe (viewed as too conservative and too low-yielding by many fixed income investors). So you are replacing two moderately risk investments with a combination of a very risky investment and a super-safe one- ie. you are barbelling your risk. Let's look at the numbers:


Returns by year (%)
Yr…...S&P500........SCV….......Bar.Agg....…Treas.….......60/40...........40/60
1991...26.0............37.5............11.0............9.9............20.0............20.9
1992...4.4............26.1............4.1............3.9............4.3...............12.8
1993...6.9............20.8............6.7............5.3............6.8...............11.5
1994...-1.4............-4.1............-5.3............-4.4............-3.0............-4.3
1995...34.6............23.0............15.4............11.6............26.9............16.2
1996...20.0............18.5............0.7............1.1............12.2............8.0
1997...30.9............29.5............7.1............5.4............21.4............15.0
1998...27.1............-8.1............7.0............7.1............19.1............1.0
1999...18.9............1.2............-3.0............-1.8............10.1............-0.6
2000...-12.4............18.5............8.0............6.9............-4.3............11.5
2001...-14.9............10.9............5.6............5.3............-6.7............7.5
2002...-23.7..........-15.8............6.7............7.7............-11.6............-1.7
2003...26.2............34.9............1.7............-0.2............16.4............13.9
2004...8.1..............20.9............1.6............-0.7............5.5............8.0
2005...1.4..............2.7............-1.0............-1.8............0.4............0.0
2006...12.4............16.0............1.0............0.3............7.9............6.6
2007...2.5............-9.9............4.1............6.0............3.2............-0.4
2008...-39.3............-34.6............1.2............12.5............-23.1............-6.4
2009...26.9............30.8............6.3............-4.7............18.7............9.5
2010...13.3............23.2............4.8............6.8............9.9............13.4


Statistics (%)
...........60/40...........40/60
mean.........6.7............7.1
st dev......12.4............7.6
worst yr...-23.1...........-6.4

So the 40/60 combo of small cap value stocks/treasuries had a better return of 7.1% vs. 6.7% for the 60/40 S&P 500/Barclays Agg Bond Index portfolio. Plus it had lower risk, standard deviation of 7.6 and worst year of -6.4% vs. a 12.4 std dev and a -23.1% worst case for the 60/40 portfolio.

So again the "barbell" strategy provided better return with lower risk.

As another example see grok's tip #7c where I demonstrate that for the years 2001-2010 a 66/34 mix of intermediate treasuries and emerging market stocks had the same return as an emerging market bond portfolio, but with lower risk (measured either by standard deviation or worst year (2008).

http://www.bogleheads.org/forum/viewtopic.php?t=69087


C) Retirement investing barbelling
See grok's tip #8 where I discuss Zvi Bodie's idea of using a combination of Long TIPs and long term equity options (LEAPS) as a better way to fund your retirement than the traditional target retirement date fund approach:

http://www.bogleheads.org/forum/viewtopic.php?t=71927

Again the barbell is between a super safe asset (long term tips which match retiree cash flow needs) and a risky asset (LEAPS).

D) Exercise Barbelling- see this recent Boglehead thread.

http://www.bogleheads.org/forum/viewtopic.php?p=843075

The idea is that a barbell of shorter periods of intense exercise and more rest is preferable to longer periods of moderate exercise.


E) Nutrition Barbelling- Based on my own extensive research I find that a meal of Fish, Vegetables and a small chocolate dessert is superior to say a comparable calorie meal of Chicken and rice with no desert! ;-)

F) You get the idea...

This is the 9th in a series of 10 investment TIPs. The series can be found here:
http://www.bogleheads.org/forum/viewtop ... highlight=

cheers,
Last edited by grok87 on Thu Apr 28, 2011 6:52 pm, edited 4 times in total.
Keep calm and Boglehead on. KCBO.

trackermike
Posts: 84
Joined: Mon Apr 05, 2010 10:33 am
Location: Upper-Midwest

Post by trackermike » Thu Apr 28, 2011 4:09 pm

I'm still digesting your post so forgive my armchair quaterbacking :) but with respect to treasuries and duration, the assumption is that there will be a parrallel shift in the yield curve. So how do you predict your strategy holds up if long rates rise more than short rates?

I applaud you for all of your research and effort on these posts.
-- | Mike | | "...no one ever had "I achieved alpha" inscribed on their tombstone." - Larry Swedroe

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Thu Apr 28, 2011 7:03 pm

trackermike wrote:I'm still digesting your post so forgive my armchair quaterbacking :) but with respect to treasuries and duration, the assumption is that there will be a parrallel shift in the yield curve. So how do you predict your strategy holds up if long rates rise more than short rates?

I applaud you for all of your research and effort on these posts.
Hi Mike,
Thanks.
Yes the numbers I gave assume a parallel yield curve shift.

There are obviously 3 possibilities for an interest rate shock:

a) parallel shift- barbelling outperforms
b) yield curve flattens- barbelling outperforms
c) yield curve steepens. For a slight steepening, barbelling might "tie" bulleting. For a large steepening barbelling would underperform.

I don't know what the historical odds are of the 3 scenarios above. My gut tells me they are all equally likely, but I don't have anything to back that up.
cheers,
Keep calm and Boglehead on. KCBO.

Choy
Posts: 352
Joined: Mon Sep 27, 2010 9:56 pm

Post by Choy » Thu Apr 28, 2011 8:59 pm

Thanks for the excellent tip!

How would you factor in TIPS in the bond portion of a portfolio? Say my fixed-income holdings are currently 50% TIPS and 50% intermediate-term treasury. Would you keep 50% in TIPS and just split the intermediate-term holding between cash and long-term so that the duration remains the same?

Follow up question -- if I wanted my bond holdings to have an average duration of 5 years, should I take into account the duration of the TIPS fund as well or should I just look at the nominal treasury funds separately?

So intermediate-term treasury fund (5 yr. duration) and TIPS fund (7.7 yr. duration) so I should split my intermediate-term treasury fund into 80% cash and 20% long-term (12.5 yr. duration) so that everything averages 5 yrs?

linuxizer
Posts: 1553
Joined: Wed Jan 02, 2008 7:55 am

Post by linuxizer » Thu Apr 28, 2011 9:34 pm

I am shocked, shocked to discover that four posts have gone in a post about barbells and bullets without my pointing out that the summary property which captures the reduced risk of a barbell here is called convexity. So I shall now rectify my failure:
A barbell has higher convexity than a portfolio with a lesser spread of durations.

Thanks grok for yet another awesome and informative post.

Bongleur
Posts: 2066
Joined: Fri Dec 03, 2010 10:36 am

Post by Bongleur » Fri Apr 29, 2011 1:09 am

Barbells bother me because you are investing in the "tails" of the distribution curve. By definition, the tails are a less statistically probable region.

And your backtest is recent. It uses a very profitable time period for bonds, where interest rates decreased steadily.

Run in from the 1940's on when bonds lost a lot of money.

Or run the recent bond data in reverse chrono order.

Its obvious where bonds are today; any tests should include if their interest rate keep increasing.

I wonder if you could count the cash twice -- cash + nominal bonds, and the same cash + TIPS. Since we can't have inflation & deflation at the same time, the cash will only come into actual play one time. Or is that stagflation -- well then count it as a 3-way... ???
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.

User avatar
Noobvestor
Posts: 4676
Joined: Mon Aug 23, 2010 1:09 am
Contact:

Post by Noobvestor » Fri Apr 29, 2011 1:13 am

What about the best of both worlds: short, intermediate AND long treasuries, all in one portfolio? :twisted:
"In the absence of clarity, diversification is the only logical strategy" -= Larry Swedroe

linuxizer
Posts: 1553
Joined: Wed Jan 02, 2008 7:55 am

Post by linuxizer » Fri Apr 29, 2011 1:21 am

Bongleur wrote:Barbells bother me because you are investing in the "tails" of the distribution curve. By definition, the tails are a less statistically probable region.

And your backtest is recent. It uses a very profitable time period for bonds, where interest rates decreased steadily.

Run in from the 1940's on when bonds lost a lot of money.

Or run the recent bond data in reverse chrono order.

Its obvious where bonds are today; any tests should include if their interest rate keep increasing.

I wonder if you could count the cash twice -- cash + nominal bonds, and the same cash + TIPS. Since we can't have inflation & deflation at the same time, the cash will only come into actual play one time. Or is that stagflation -- well then count it as a 3-way... ???
He's assuming a parallel yield curve shift. So it's making no assumptions about the stability of the tails, it's simply what happens when a portfolio with positive convexity experiences an across-the-board interest rate change of any kind: the convexity improves the outcome of the rate change in either direction on the present value of the portfolio.

User avatar
Daffy
Posts: 152
Joined: Tue Nov 06, 2007 4:05 pm

Re: Grok's Tip#9: To Manage Risk Well, Use a Bar-Bell

Post by Daffy » Fri Apr 29, 2011 2:48 am

grok87 wrote: Small Cap Value stocks are generally viewed as more risky equities (the market hates those companies, hence their low market cap and valuation)
How can you write such drivel and have such a loyal following in this forum? I don't get it. I never knew that because the "market hates" small companies that is the reason for their "low market cap and valuation."

User avatar
fredflinstone
Posts: 2096
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone » Fri Apr 29, 2011 6:22 am

Thank you for a very interesting post.

Your bond barbell idea sounds very convincing, but since the treasury bond market is super-liquid, I can't believe the market has somehow overlooked this seeming anomaly. If your strategy was as much a sure thing as you seem to suggest, wouldn't hedge funds use arbitrage to eliminate any advantage to it?

For example, couldn't John Paulson simply buy $50 billion worth of the t-bills and 7-year treasury bonds and then simultaneously short $50 billion worth of 3-year treasury bonds? By your accounting, this should be an almost sure bet regardless of what happens to interest rates.

(Or, since Mr. Paulson likes leverage, he could implement a turbo-boosted version of the same strategy by buying calls on t-bills and 7-year treasuries while simultaneously buying puts on 3-year treasuries.)

Sorry, I don't believe in free lunches. There must be an element of risk here that you are not accounting for properly, e.g., perhaps the risk of a steepening yield curve is higher than you think.

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Fri Apr 29, 2011 7:10 am

Choy wrote:Thanks for the excellent tip!

How would you factor in TIPS in the bond portion of a portfolio? Say my fixed-income holdings are currently 50% TIPS and 50% intermediate-term treasury. Would you keep 50% in TIPS and just split the intermediate-term holding between cash and long-term so that the duration remains the same?

Follow up question -- if I wanted my bond holdings to have an average duration of 5 years, should I take into account the duration of the TIPS fund as well or should I just look at the nominal treasury funds separately?

So intermediate-term treasury fund (5 yr. duration) and TIPS fund (7.7 yr. duration) so I should split my intermediate-term treasury fund into 80% cash and 20% long-term (12.5 yr. duration) so that everything averages 5 yrs?
Hi Choy. Thanks, glad it was helpful.
Interesting question. I guess I would barbell the treasuries and Tips separtely and try to match their respective durations.
When you barbell the treasuries you can use an fdic insured savings acct earning 1% for the cash part. When you barbell the tips you can used i-bonds for the cash part. Let's look at the tips barbell a little more closely. I'm going to use average maturity instead of duration because duration can be a little confusing for tips- (there's duration with respect to nominal rates and duration with respect to real rates).
The vanguard tips fund has an average maturity of about 9 years and a real yield of 0% (shocking but true!)
So instead you could go with a barbell of 70% In ibonds and 30% in 30 year tips (which have a real yield of 1.75%). That would give you a real yield of 0.53%, which is a whole lot better than 0. Plus the ibonds give you a cheap put option. If rates rise dramatically you can cash out at close to par and reinvest at the higher rates.
Whereas a tips fund would take a big market value hit from the rate rise.
Cheers,
Keep calm and Boglehead on. KCBO.

linuxizer
Posts: 1553
Joined: Wed Jan 02, 2008 7:55 am

Post by linuxizer » Fri Apr 29, 2011 7:41 am

fredflinstone wrote:Your bond barbell idea sounds very convincing, but since the treasury bond market is super-liquid, I can't believe the market has somehow overlooked this seeming anomaly. If your strategy was as much a sure thing as you seem to suggest, wouldn't hedge funds use arbitrage to eliminate any advantage to it?
There's no contradiction. There is a price to convexity in the market, plain and simple. As usual, the easiest way to think about convexity is with MBSs: the yield on a GNMA fund is higher than a similar Treasury fund because of the negative convexity; but the same is true to a lesser degree for the smaller differences between a portfolio with moderate positive convexity vs. one with a little more positive convexity.

This is one reason why I have asked Vanguard to take the lead in making their funds' convexity more transparent and publishing it alongside duration and other information--that it is easy for unscrupulous fund managers scrambling for every .01% of yield to trade off some convexity for that little bit of yield. As a firm that I assume doesn't cheat in this way, it would be in Vanguard's best interest to take the lead in getting this information out there, but they don't seem to be.

Choy
Posts: 352
Joined: Mon Sep 27, 2010 9:56 pm

Post by Choy » Fri Apr 29, 2011 8:43 am

grok87 wrote:Hi Choy. Thanks, glad it was helpful.
Interesting question. I guess I would barbell the treasuries and Tips separtely and try to match their respective durations.
When you barbell the treasuries you can use an fdic insured savings acct earning 1% for the cash part. When you barbell the tips you can used i-bonds for the cash part. Let's look at the tips barbell a little more closely. I'm going to use average maturity instead of duration because duration can be a little confusing for tips- (there's duration with respect to nominal rates and duration with respect to real rates).
The vanguard tips fund has an average maturity of about 9 years and a real yield of 0% (shocking but true!)
So instead you could go with a barbell of 70% In ibonds and 30% in 30 year tips (which have a real yield of 1.75%). That would give you a real yield of 0.53%, which is a whole lot better than 0. Plus the ibonds give you a cheap put option. If rates rise dramatically you can cash out at close to par and reinvest at the higher rates.
Whereas a tips fund would take a big market value hit from the rate rise.
Cheers,
Thanks for the follow-up grok, your suggestion of splitting the two and using cash and ibonds sounds very interesting.

I'll be sure to keep this strategy in mind as I research more about bonds and how they work in various situations. For such a seemingly safe and "boring" investment, bonds are far more complicated to me than stocks!

User avatar
DartThrower
Posts: 804
Joined: Wed Mar 11, 2009 4:10 pm
Location: Philadelphia

Post by DartThrower » Fri Apr 29, 2011 10:16 am

linuxizer wrote:
fredflinstone wrote:Your bond barbell idea sounds very convincing, but since the treasury bond market is super-liquid, I can't believe the market has somehow overlooked this seeming anomaly. If your strategy was as much a sure thing as you seem to suggest, wouldn't hedge funds use arbitrage to eliminate any advantage to it?
There's no contradiction. There is a price to convexity in the market, plain and simple. As usual, the easiest way to think about convexity is with MBSs: the yield on a GNMA fund is higher than a similar Treasury fund because of the negative convexity; but the same is true to a lesser degree for the smaller differences between a portfolio with moderate positive convexity vs. one with a little more positive convexity.

This is one reason why I have asked Vanguard to take the lead in making their funds' convexity more transparent and publishing it alongside duration and other information--that it is easy for unscrupulous fund managers scrambling for every .01% of yield to trade off some convexity for that little bit of yield. As a firm that I assume doesn't cheat in this way, it would be in Vanguard's best interest to take the lead in getting this information out there, but they don't seem to be.
Hi linuxizer.

These would have been great issues to bring up at our recent local Philly chapter meeting where we discussed bonds in detail, but of course the meeting was only 2 hours. :) If we had more time we could have looked at the kinds of topics covered in this post in detail. Nevertheless the meeting was very valuable in teaching about the basics of duration and convexity in both bond funds and individually constructed bond portfolios. We all very much appreciate your efforts at that meeting to explain the concepts and to illustrate them to us as well.

With regard to your efforts to get Vanguard to take the lead in educating the bond fund investors about convexity, we can bring this up at the next Boglehead reunion in the Fall. There is a great variation in the level of sophistication among Vanguard investors. While I agree with you about Vanguard getting out the information, my guess is that Vanguard does not want to confuse the less sophisticated investors or those without the time to carefully consider the implications of convexity in their portfolios. So the real trick is to get the word out in a way that is effective for a broad cross section of Vanguard investors.
A Boglehead can stay the course longer than the market can stay irrational. It's impossible to have a bubble in common sense.

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Fri Apr 29, 2011 10:51 am

linuxizer wrote:I am shocked, shocked to discover that four posts have gone in a post about barbells and bullets without my pointing out that the summary property which captures the reduced risk of a barbell here is called convexity. So I shall now rectify my failure:
A barbell has higher convexity than a portfolio with a lesser spread of durations.

Thanks grok for yet another awesome and informative post.
Linuxizer,
Your welcome and thanks for the link to the wik article on convexity. The wiki article is good. I do have slight quibble with this sentence though
Bogleheads wiki on convexity wrote: Simple bonds, with no call or prepayment features, have slightly positive convexity because of the way duration works out.
the quibble is that for some "simple" bonds the positive convexity can be pretty material- i.e. more than just "slightly positive." This is a good thing from a risk perspective. An example would be the 30 year treasury. I'll try to post some numbers to illustrate this when i get a chance...
cheers,
Keep calm and Boglehead on. KCBO.

User avatar
Mian
Posts: 92
Joined: Wed Jun 03, 2009 6:02 am

Post by Mian » Fri Apr 29, 2011 10:52 am

Thanks for #9 Grok!

thehammer
Posts: 138
Joined: Tue Dec 28, 2010 3:15 am

Post by thehammer » Fri Apr 29, 2011 11:30 am

it's weird, is this an unusual or usual, let's look at treasury rates right now


3 Month 0.02 0.01 0.03
6 Month 0.08 0.08 0.09
2 Year 0.61 0.62 0.65
3 Year 1.01 1.02 1.15
5 Year 1.98 2.00 2.11
10 Year 3.29 3.31 3.40
30 Year 4.39 4.41 4.47

First column is current yield. 5 YEAR yield is higher than taking half of the 3 month and 10 year. 1.65 vs 1.90

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Fri Apr 29, 2011 12:50 pm

Bongleur wrote:Barbells bother me because you are investing in the "tails" of the distribution curve. By definition, the tails are a less statistically probable region.

And your backtest is recent. It uses a very profitable time period for bonds, where interest rates decreased steadily.

Run in from the 1940's on when bonds lost a lot of money.

Or run the recent bond data in reverse chrono order.

Its obvious where bonds are today; any tests should include if their interest rate keep increasing.

I wonder if you could count the cash twice -- cash + nominal bonds, and the same cash + TIPS. Since we can't have inflation & deflation at the same time, the cash will only come into actual play one time. Or is that stagflation -- well then count it as a 3-way... ???
Hi Bongleur,
I'm not quite following your logic. I assume your talking about Part A) of my post. There is no backtest in that anywhere. It a purely prospective look at future expected return and risk for different bond strategies. If you are looking for where I get my numbers from (the 10%, 11% etc.) I could walk you through how to do it in Excel.

Re your comment on the "tails": when you barbell you are actually positioning yourself to outperform in the tail scenarios. For example the 400 bp interest rate shock is a tail scenario- i.e. it is thought by most to be rather unlikely. The good news is that the Fed is finally forcing banks to consider these type of scenarios in their risk management. I think individual investors should consider them too. People are used to (or should be used to) doing this on the equity side. A common rulle of thumb is to plan that your equities could lose half their value. People are not so used to doing this on the bond side, because, as you point out it has been a very benign environment for bonds in recent decades. The starting point on the bond side is to plan that your bonds or bond funds could lose 4*duration. i.e. if you have a bond fund with a 5 year duration, it might lose as much as 20%
cheers,
Keep calm and Boglehead on. KCBO.

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Fri Apr 29, 2011 12:58 pm

fredflinstone wrote:Thank you for a very interesting post.

Your bond barbell idea sounds very convincing, but since the treasury bond market is super-liquid, I can't believe the market has somehow overlooked this seeming anomaly. If your strategy was as much a sure thing as you seem to suggest, wouldn't hedge funds use arbitrage to eliminate any advantage to it?

For example, couldn't John Paulson simply buy $50 billion worth of the t-bills and 7-year treasury bonds and then simultaneously short $50 billion worth of 3-year treasury bonds? By your accounting, this should be an almost sure bet regardless of what happens to interest rates.

(Or, since Mr. Paulson likes leverage, he could implement a turbo-boosted version of the same strategy by buying calls on t-bills and 7-year treasuries while simultaneously buying puts on 3-year treasuries.)

Sorry, I don't believe in free lunches. There must be an element of risk here that you are not accounting for properly, e.g., perhaps the risk of a steepening yield curve is higher than you think.
Your welcome fred.
Interesting points. Note I never said it was an "almost sure bet". I said the expected return of the tbill and 7 year treasury portfolio was higher than the 3 year treasury.
Normally, as linuxizer points out, when you barbell you get a slightly lower yield in exchange for the higher positive convexity (the higher positive convexity lowers your risk from a largish interest rate shock). I don't know why that's not true for the example I gave. It may have something to do with QE2 perhaps?
cheers,
Keep calm and Boglehead on. KCBO.

thehammer
Posts: 138
Joined: Tue Dec 28, 2010 3:15 am

Post by thehammer » Fri Apr 29, 2011 1:31 pm

grok is why the 5 year yielding substantially more than the 0/10 year average right now

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Fri Apr 29, 2011 1:50 pm

thehammer wrote:grok is why the 5 year yielding substantially more than the 0/10 year average right now
hi thehammer,
I'm not sure but its not entirely unexpected. Normally there is a cost for barbelling. I.e. you give up a bit in yield and expected return in exchange for the greater protection in an interest rate shock scenario.
Keep calm and Boglehead on. KCBO.

linuxizer
Posts: 1553
Joined: Wed Jan 02, 2008 7:55 am

Post by linuxizer » Fri Apr 29, 2011 4:19 pm

grok - I agree with your quibble, plus the second half of that sentence is horribly vague. Fixed.

DartThrower - Hopefully I gave everyone tools to model these portfolios for themselves! :-)
But convexity is fascinating. Maybe we'll talk about it more in a future meeting.

User avatar
Lbill
Posts: 4997
Joined: Thu Mar 13, 2008 11:25 pm
Location: Somewhere between Up and Down

Post by Lbill » Fri Apr 29, 2011 5:53 pm

Grok - I tried this exercise with IShares Treasury bond ETFs. I used SHV (0.37 duration) with IEF, TLH, and TLT - weighting the barbell to achieve the averaged duration for IEI (4.43 yrs duration). Using the SEC 30-day yields that are given for each ETF, I found that every single bond barbell that matched the average duration of IEI had an average SEC 30-day yield that was less than the 30-day yield for IEI. Perhaps you can share the details of how you would do these calculations, or check mine. Unless I'm in error, your bond barbell strategy doesn't currently work using these Treasury ETFs. The average SEC 30-day yields for barbells matching the average duration of IEI are all lower than simply investing 100% in IEI.
"Life can only be understood backward; but it must be lived forward." ~ Søren Kierkegaard | | "You can't connect the dots looking forward; but only by looking backwards." ~ Steve Jobs

linuxizer
Posts: 1553
Joined: Wed Jan 02, 2008 7:55 am

Post by linuxizer » Fri Apr 29, 2011 5:55 pm

Lbill wrote:Grok - I tried this exercise with IShares Treasury bond ETFs. I used SHV (0.37 duration) with IEF, TLH, and TLT - weighting the barbell to achieve the averaged duration for IEI (4.43 yrs duration). Using the SEC 30-day yields that are given for each ETF, I found that every single bond barbell that matched the average duration of IEI had an average SEC 30-day yield that was less than the 30-day yield for IEI. Perhaps you can share the details of how you would do these calculations, or check mine. Unless I'm in error, your bond barbell strategy doesn't currently work using these Treasury ETFs. The average SEC 30-day yields for barbells matching the average duration of IEI are all lower than simply investing 100% in IEI.
It seems to me that you're proving the point. Convexity ("barbellness"!) costs yield in an efficient market, because it is valuable.

Bongleur
Posts: 2066
Joined: Fri Dec 03, 2010 10:36 am

Post by Bongleur » Fri Apr 29, 2011 8:08 pm

>
People are used to (or should be used to) doing this on the equity side. A common rulle of thumb is to plan that your equities could lose half their value. People are not so used to doing this on the bond side, because, as you point out it has been a very benign environment for bonds in recent decades. The starting point on the bond side is to plan that your bonds or bond funds could lose 4*duration. i.e. if you have a bond fund with a 5 year duration, it might lose as much as 20%
>

Why 4 times ?
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Fri Apr 29, 2011 8:16 pm

Lbill wrote:Grok - I tried this exercise with IShares Treasury bond ETFs. I used SHV (0.37 duration) with IEF, TLH, and TLT - weighting the barbell to achieve the averaged duration for IEI (4.43 yrs duration). Using the SEC 30-day yields that are given for each ETF, I found that every single bond barbell that matched the average duration of IEI had an average SEC 30-day yield that was less than the 30-day yield for IEI. Perhaps you can share the details of how you would do these calculations, or check mine. Unless I'm in error, your bond barbell strategy doesn't currently work using these Treasury ETFs. The average SEC 30-day yields for barbells matching the average duration of IEI are all lower than simply investing 100% in IEI.
Lbill, i agree with linuxizer's post above. The primary purpose of barbelling is to manage or reduce interest rate risk. Sometimes you get a better yield as in the examples I gave. But not always. Why not do your barbell using an FDIC insured savings account yielding 1%? I bet you'll get a better yield then...
Keep calm and Boglehead on. KCBO.

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Fri Apr 29, 2011 8:38 pm

Bongleur wrote:>
People are used to (or should be used to) doing this on the equity side. A common rulle of thumb is to plan that your equities could lose half their value. People are not so used to doing this on the bond side, because, as you point out it has been a very benign environment for bonds in recent decades. The starting point on the bond side is to plan that your bonds or bond funds could lose 4*duration. i.e. if you have a bond fund with a 5 year duration, it might lose as much as 20%
>

Why 4 times ?
the 4 times comes from the 400 bp interest rate shock scenario. In that scenario, for example, the 10 year treasury yield would go from 3.3% (today) to 7.3%.
Keep calm and Boglehead on. KCBO.

User avatar
Lbill
Posts: 4997
Joined: Thu Mar 13, 2008 11:25 pm
Location: Somewhere between Up and Down

Post by Lbill » Fri Apr 29, 2011 8:42 pm

Grok - I was looking for a free lunch with the barbell. I guess it's more realistic to expect a smaller lunch that's less likely to be interrupted by a noisy table of diners nearby?
"Life can only be understood backward; but it must be lived forward." ~ Søren Kierkegaard | | "You can't connect the dots looking forward; but only by looking backwards." ~ Steve Jobs

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Fri Apr 29, 2011 9:10 pm

Lbill wrote:Grok - I was looking for a free lunch with the barbell. I guess it's more realistic to expect a smaller lunch that's less likely to be interrupted by a noisy table of diners nearby?
ok here's the free lunch:

a) IEI: Yield = 2.08%, duration =4.46

b)
62% IEF Yield = 3.16%, duration =7.28
38% FDIC Insured Saving acct., yield =1%, duration =0
combo: yield =2.34%, duration =4.5

cheers,
Keep calm and Boglehead on. KCBO.

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Fri Apr 29, 2011 9:14 pm

Noobvestor wrote:What about the best of both worlds: short, intermediate AND long treasuries, all in one portfolio? :twisted:
I guess I'm not a fan of intermediate bonds these days. For example the 3 year treasury is yielding 1% but has interest rate risk. You can get 1% from a FDIC insured savings account, without any interest rate risk. So why take risk when you are not getting paid to do so?
cheers,
Keep calm and Boglehead on. KCBO.

User avatar
Lbill
Posts: 4997
Joined: Thu Mar 13, 2008 11:25 pm
Location: Somewhere between Up and Down

Post by Lbill » Fri Apr 29, 2011 9:48 pm

Grok - now the only rub is how I can get that 1% FDIC insured savings account inside my IRA where IEF currently resides.

What do you think about taking a smaller position in the long Treasury ETF (TLT) to get the same averaged duration? How would one decide between the Treasury ETFs with different durations? This is beginning to sound like Harry Browne's Permanent Portfolio which uses equal portions of short and long Treasuries in a bond barbell. Maybe he was onto something.
"Life can only be understood backward; but it must be lived forward." ~ Søren Kierkegaard | | "You can't connect the dots looking forward; but only by looking backwards." ~ Steve Jobs

User avatar
wintermute
Posts: 188
Joined: Mon Mar 15, 2010 10:36 pm

Post by wintermute » Sat Apr 30, 2011 12:00 am

Lbill wrote:Grok - now the only rub is how I can get that 1% FDIC insured savings account inside my IRA where IEF currently resides.
Penfed has IRA CD's. But then if you need to use it rebalance into equities, you'd have to deal with closing the CD (discounting the penalty) and wait for the transfer to complete. Depending on the portfolio's AA and size, it would make sense to put some of the short end in an IRA CD.

User avatar
LadyGeek
Site Admin
Posts: 46055
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Post by LadyGeek » Sat Apr 30, 2011 4:32 am

Let me bring some prior thoughts into the mix, circa 2000. Start with Bill Bernstein's Of Markets and Barbells (a good tutorial). Then, read the follow-up discussion in Morningstar: Of Barbells and Markets (M*), which included opinions by Bill Bernstein (wbern) and Larry Swedroe. Do these opinions still apply 11 years later?

Reference - Wiki article link: Index to important Posts on Morningstar

================
I was also at the Philly chapter meeting with Linuxizer and DartThrower. As Linuxizer stated previously, fund managers are scrambling for every .01% of yield. The other half of the statement was that convexity has very little impact to the average investor - you need to be working in large $$$ to see the effects.

Advanced math is needed to understand convexity in detail. Linuxizer's wiki reference alludes to this - Investopedia avoids the math details totally, while Wikipedia has no graphs; an example of 2 extremes. It's a good thing I brought my textbook to the meeting. Linuxizer's software is under development in "R" (RStudio for the graphic interface version).
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

linuxizer
Posts: 1553
Joined: Wed Jan 02, 2008 7:55 am

Post by linuxizer » Sat Apr 30, 2011 5:52 am

LadyGeek wrote:The other half of the statement was that convexity has very little impact to the average investor - you need to be working in large $$$ to see the effects.
As grok pointed out thought, if convexity is large enough it can make a noticeable difference.
Advanced math is needed to understand convexity in detail. Linuxizer's wiki reference alludes to this - Investopedia avoids the math details totally, while Wikipedia has no graphs; an example of 2 extremes. It's a good thing I brought my textbook to the meeting. Linuxizer's software is under development in "R" (RStudio for the graphic interface version).
The math isn't terribly bad to understand it, but since it's a squared term interpreting it via rules-of-thumb is a little non-intuitive.

For anyone interested in the software, I'm going to do some bug-fixing based on comments in the chapter meeting and release it soon.

User avatar
Doc
Posts: 8293
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Post by Doc » Sat Apr 30, 2011 8:41 am

Lbill wrote:Grok - now the only rub is how I can get that 1% FDIC insured savings account inside my IRA where IEF currently resides.
With only a 1% return the tax loss is small even for high tax rate investors (35 bps) and neither of the options should be taking up that valuable tax advantaged space.

User avatar
Lbill
Posts: 4997
Joined: Thu Mar 13, 2008 11:25 pm
Location: Somewhere between Up and Down

Post by Lbill » Sat Apr 30, 2011 9:28 am

From LadyGeek:
Let me bring some prior thoughts into the mix, circa 2000. Start with Bill Bernstein's Of Markets and Barbells (a good tutorial). Then, read the follow-up discussion in Morningstar: Of Barbells and Markets (M*), which included opinions by Bill Bernstein (wbern) and Larry Swedroe. Do these opinions still apply 11 years later?
Having just re-read this, it seems the conclusion is that barbelling between large and small/microcap doesn't get you anything - especially within the context of a port than includes bonds. TSM + Bonds gets you to the same place as LC+SC+Bonds. Interesting. Does this imply that it won't get you much to add SC to TSM in your porfolio (might as well just stick with TSM)? I wonder if this would apply to an international small cap vs. large cap barbell as well? In that case, adding VSS (int small cap) to VEU (int TSM) won't get you anything either?
"Life can only be understood backward; but it must be lived forward." ~ Søren Kierkegaard | | "You can't connect the dots looking forward; but only by looking backwards." ~ Steve Jobs

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Sat Apr 30, 2011 11:44 am

Doc wrote:
Lbill wrote:Grok - now the only rub is how I can get that 1% FDIC insured savings account inside my IRA where IEF currently resides.
With only a 1% return the tax loss is small even for high tax rate investors (35 bps) and neither of the options should be taking up that valuable tax advantaged space.
Good point Doc. I hold mine in taxable for that reason. I just think of it as a muni money market with a better yield and without the risk.
My 401k/ira space is all taken up with domestic and foreign reits, very long term tips, and credit union cds.
Cheers,
Keep calm and Boglehead on. KCBO.

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Post by grok87 » Mon May 02, 2011 8:17 am

LadyGeek wrote:Let me bring some prior thoughts into the mix, circa 2000. Start with Bill Bernstein's Of Markets and Barbells (a good tutorial). Then, read the follow-up discussion in Morningstar: Of Barbells and Markets (M*), which included opinions by Bill Bernstein (wbern) and Larry Swedroe. Do these opinions still apply 11 years later?

Reference - Wiki article link: Index to important Posts on Morningstar

================
I was also at the Philly chapter meeting with Linuxizer and DartThrower. As Linuxizer stated previously, fund managers are scrambling for every .01% of yield. The other half of the statement was that convexity has very little impact to the average investor - you need to be working in large $$$ to see the effects.

Advanced math is needed to understand convexity in detail. Linuxizer's wiki reference alludes to this - Investopedia avoids the math details totally, while Wikipedia has no graphs; an example of 2 extremes. It's a good thing I brought my
textbook to the meeting. Linuxizer's software is under development in "R" (RStudio for the graphic interface version).
Thanks ladygeek,
I skimmed the previous threads. Agree with lbill that the barbell being discussed seems mostly about using large and small cap stocks and skipping midcaps.
Also the discussion seems to be about 11 yeArs old
Investors have a right to be skeptical of backtested data. Perhaps the case for a treasury-small value stock barbell is spurious and rests on anomalous period? Is there any intuitive reason why it would make sense?
I think there is and that it boils down to inflation-deflation.
Small value companies are fragile highly leveraged companies with lots of debt. If there is unexpected inflation they will benefit as their debt burden will be reduced in real terms. Treasuries on the other hand will be hurt by the unexpected inflation. If there is deflation ala the great depression the opposite will happen- small value will be crushed but treasuries wil be king.
So the two make a great pairing.
I'll try to run the numbers on the great depression and the great inflation periods when I get a chance
Cheers,
Keep calm and Boglehead on. KCBO.

cosmicdebris
Posts: 2
Joined: Tue Jun 07, 2011 8:03 pm

Barbelling TSP

Post by cosmicdebris » Tue Jun 07, 2011 8:11 pm

I have a question about using the barbell technique for a TSP account.

Let's assume I'm going to use three of the TSP funds: G, C, and S (international is covered in taxable with VGTSX and other bonds are covered in my Roth with VBMFX).

Would the "best" allocation be:

A: 60/40 with C/S matching TSM

B: 50/50 with a C/S split

or C: 40/60 with all or mostly S

I'm currently closest to B but am leaning toward C. Does this fit the barbell philosophy or am I missing something? (thanks)

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Re: Barbelling TSP

Post by grok87 » Tue Jun 07, 2011 9:38 pm

cosmicdebris wrote:I have a question about using the barbell technique for a TSP account.

Let's assume I'm going to use three of the TSP funds: G, C, and S (international is covered in taxable with VGTSX and other bonds are covered in my Roth with VBMFX).

Would the "best" allocation be:

A: 60/40 with C/S matching TSM

B: 50/50 with a C/S split

or C: 40/60 with all or mostly S

I'm currently closest to B but am leaning toward C. Does this fit the barbell philosophy or am I missing something? (thanks)
hello cosmicdebris and welcome to the forum,
the S fund isn't quite the small value fund I mentioned. And while I like the G fund a whole lot (the yield of long term treasuries without the interest rate risk) it isn't the treasury fund from my example either (won't give you a boost like long term treasuries did in 2008).

The closest thing in your situation might be this

since you can use the G fund for your treasuries you could go longer duration with your TIPs (do you own these elsewhere?)
So that is in effect a barbell of short duration (but high yielding) nominal treasuries and longer duration TIPs.

cheers,
Keep calm and Boglehead on. KCBO.

cosmicdebris
Posts: 2
Joined: Tue Jun 07, 2011 8:03 pm

Re: Barbelling TSP

Post by cosmicdebris » Wed Jun 08, 2011 5:09 pm

Thanks grok. I do have a bit of VIPSX but the duration in there is only five years or so, I think. I was just thinking that while the S fund is not small value it may have enough risk/reward expectations vs. the C fund that it could be used in a similar manner for barbelling since the G fund is ultra-safe.

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

tip 10

Post by grok87 » Fri Oct 14, 2011 9:55 am

Here's tip 10: Get Real (Returns)!- the 3% solution

http://www.bogleheads.org/forum/viewtop ... highlight=

cheers,
Keep calm and Boglehead on. KCBO.

User avatar
BlueEars
Posts: 3599
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Post by BlueEars » Fri Oct 14, 2011 10:42 am

LadyGeek wrote:Let me bring some prior thoughts into the mix, circa 2000. Start with Bill Bernstein's Of Markets and Barbells (a good tutorial). Then, read the follow-up discussion in Morningstar: Of Barbells and Markets (M*), which included opinions by Bill Bernstein (wbern) and Larry Swedroe. Do these opinions still apply 11 years later?
...
I wonder about that too. Here are the stats for MSCI small, mid, large caps as used now by Vanguard funds. CAGR's are for Jan-2000 to Sept-2011.

Code: Select all

SV 8.0%
SG 2.6&
MV 8.1%
MG 1.0%
LV 2.6%
LG 1.0%
Midcaps did OK for that time period and MV slightly beat SV. Also MV had stdev=0.052 whereas SV had stdev=0.056. MSCI's SV is not as small-value as DFA's offering so maybe the story is a little different at DFA?

User avatar
LH
Posts: 5490
Joined: Wed Mar 14, 2007 2:54 am

Re: Grok's Tip#9: To Manage Risk Well, Use a Bar-Bell

Post by LH » Fri Oct 14, 2011 1:17 pm

Daffy wrote:
grok87 wrote: Small Cap Value stocks are generally viewed as more risky equities (the market hates those companies, hence their low market cap and valuation)
How can you write such drivel and have such a loyal following in this forum? I don't get it. I never knew that because the "market hates" small companies that is the reason for their "low market cap and valuation."
Its not drivel. If you "don't get it", why call it "drivel"? Why insult?

To answer your quesiton, and assume you are not just trolling, its just a shorthand way, of talking about French Fama, and the purported small cap value premium, with the use of "hates" an implicit reference to the behavorial economics interpretation of the purported SCV premium. FF as I understand it, say its due to higher "risk" that SCV will expectationally outperfrom say TSM. The behavoiralists, say that humans like "sexy" growth stocks more, (and always will?? I dunno the exact position), and that human behavioral shortcomings, explain, some (most/all?) off the premium. Anyway, when I read that, thats what I "thought", just seemed par for the course. I am not saying its true, its just a position in valid contention, that is well known. He cannot avoid shorthand, without becoming long winded, and boring others to tears.

If you do not know, just ask. Please do not call things "drivel" even if you think you know its drivel for sure. Its highly meaningless, anyone can insult, and its just impolite, and just because you think somethings drivel, does not mean it is.

thanks,

LH

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

tip 11 is out

Post by grok87 » Sun Dec 18, 2011 10:47 pm

Keep calm and Boglehead on. KCBO.

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Re: Grok's Tip#9: To Manage Risk Well, Use a Bar-Bell

Post by grok87 » Tue Oct 16, 2012 10:39 pm

Keep calm and Boglehead on. KCBO.

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Re: Grok's Tip#9: To Manage Risk Well, Use a Bar-Bell

Post by grok87 » Tue Nov 20, 2012 7:45 am

Keep calm and Boglehead on. KCBO.

User avatar
archbish99
Posts: 1632
Joined: Fri Jun 10, 2011 6:02 pm

Re: Grok's Tip#9: To Manage Risk Well, Use a Bar-Bell

Post by archbish99 » Wed Nov 21, 2012 8:32 pm

Admittedly, an old one, but I'm confused exactly how this works. Let's say I wanted to barbell bond funds. I could either go with IT Bond Index (1.73% yield, 6.4 year duration), or I could go with a mix of 31% LT Bond Index (3.39% yield, 14.8 year duration) and 69% ST Bond Index (0.53% yield, 2.7 year duration). The mix has a duration of 6.4 years, and a yield of 1.42%.

Alternatively, I could use a 43% LTBI and 57% Prime Money Market (0.03% yield, zero duration) barbell and get a yield of 1.47% with equal duration. It feels like your requirement here is to have a zero-duration asset which still has a decently positive yield.

Regardless, since the duration is the same, wouldn't that mean the net effect of an interest rate shock is also the same? A 4% increase in rates would drop ITBI by 25.6%, or would drop the barbell by (31% * 59.2%) + (69% * 10.8%) = 25.8% -- effectively the same. I'm having trouble seeing what the benefit is here for which I would be paying 26-31bp of yield.
I'm not a financial advisor, I just play one on the Internet.

grok87
Posts: 8114
Joined: Tue Feb 27, 2007 9:00 pm

Re: Grok's Tip#9: To Manage Risk Well, Use a Bar-Bell

Post by grok87 » Wed Nov 21, 2012 11:26 pm

archbish99 wrote:Admittedly, an old one, but I'm confused exactly how this works. Let's say I wanted to barbell bond funds. I could either go with IT Bond Index (1.73% yield, 6.4 year duration), or I could go with a mix of 31% LT Bond Index (3.39% yield, 14.8 year duration) and 69% ST Bond Index (0.53% yield, 2.7 year duration). The mix has a duration of 6.4 years, and a yield of 1.42%.

Alternatively, I could use a 43% LTBI and 57% Prime Money Market (0.03% yield, zero duration) barbell and get a yield of 1.47% with equal duration. It feels like your requirement here is to have a zero-duration asset which still has a decently positive yield.

Regardless, since the duration is the same, wouldn't that mean the net effect of an interest rate shock is also the same? A 4% increase in rates would drop ITBI by 25.6%, or would drop the barbell by (31% * 59.2%) + (69% * 10.8%) = 25.8% -- effectively the same. I'm having trouble seeing what the benefit is here for which I would be paying 26-31bp of yield.
Hi archbish,
well let's start with your second barbell and replace the prime money market with say an FDIC insured savings account. You should be able to get 0.5% or so without much effort. That would bring the yield of the barbell up to the same 1.74% or so of the IT Bond index for the same duration.
Now the benefit comes in the form of the higher (better) convexity. Vanguard doesn't provide the convexity numbers (because they are not required to and they probably feel it will confuse people). But the higher the convexity the better protection you have against large interest rate shocks. From the Barclays website the convexity figures (for the indexes, not the funds) are:

intermediate bond index 0.47
long term bond index 3.15

the convexity of cash/tbills is 0.

so the convexity of the 57% fdic insured savings acct./43% long term bond index is 1.35 which is quite a bit higher than the 0.47 convexity of the intermediate bond index.


here's a short example to show how convexity works. let's say a bond fund has a duration of 10 and a convexity of 2. Then let's say interest rates go up 3% points (i.e. 300 bps). The simple duration approximation formula is inaccurate because of the large size of the interest rate shock. The simple duration approximation estimates that the bond fund's price will change by -30% ( = -10*3)/100. The duration/convexity approximation estimates that the bonds fund price will change by -21% ( = -10*3 + 0.5*2*3*3)/100 ). So for large interest rate shocks, convexity becomes very material and you want as much as possible of it to protect yourself.
cheers,
Keep calm and Boglehead on. KCBO.

User avatar
archbish99
Posts: 1632
Joined: Fri Jun 10, 2011 6:02 pm

Re: Grok's Tip#9: To Manage Risk Well, Use a Bar-Bell

Post by archbish99 » Fri Nov 23, 2012 8:25 am

grok87 wrote:Now the benefit comes in the form of the higher (better) convexity. Vanguard doesn't provide the convexity numbers (because they are not required to and they probably feel it will confuse people). But the higher the convexity the better protection you have against large interest rate shocks. From the Barclays website the convexity figures (for the indexes, not the funds) are:

intermediate bond index 0.47
long term bond index 3.15

the convexity of cash/tbills is 0.

so the convexity of the 57% fdic insured savings acct./43% long term bond index is 1.35 which is quite a bit higher than the 0.47 convexity of the intermediate bond index.


here's a short example to show how convexity works. let's say a bond fund has a duration of 10 and a convexity of 2. Then let's say interest rates go up 3% points (i.e. 300 bps). The simple duration approximation formula is inaccurate because of the large size of the interest rate shock. The simple duration approximation estimates that the bond fund's price will change by -30% ( = -10*3)/100. The duration/convexity approximation estimates that the bonds fund price will change by -21% ( = -10*3 + 0.5*2*3*3)/100 ). So for large interest rate shocks, convexity becomes very material and you want as much as possible of it to protect yourself.
Ah -- this is the point I was missing: The extended formula for a bond's change based on duration*change and 1/2*convexity*(change^2). So it's fair to say that convexity buffers rate increases, and thus equivalent duration with better convexity is a good thing. Thanks for the clarification.
I'm not a financial advisor, I just play one on the Internet.

Post Reply