A simulated historical TIPS data set...

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
Topic Author
Cb
Posts: 274
Joined: Wed Feb 21, 2007 12:39 am

A simulated historical TIPS data set...

Post by Cb » Wed Feb 28, 2007 1:18 pm

A number of weeks ago like many of you I was copying & pasting TrevH's data on a variety of asset class returns from 1972 thru 2006. I then began to search for information on TIPS behavior. Time and again this Kothari paper was mentioned (most recently by Larry Swedroe on a DH thread) as having done a credible job of constructing a simulated TIPS dataset starting in 1953:

http://gnobility.com/ER/ShankenKothari_ ... h_TIPS.pdf

I emailed Dr. Kothari at MIT and asked him the following:

Dr. Kothari,

I'm, a recently-retired individual investor trying to improve my knowledge of investing. As my wife and I roll-out our 401K's to Vanguard we're going to a more broadly diversified portfolio using MPT concepts.

We now have access to a low cost cost TIPS fund at Vanguard and I'm giving seriously thought to diversifying our bond portfolio with TIPS. As I look for historical data on inflation-protected securities your name pops up frequently, most often mentioning a 2004 paper you had published in the Financial Analysts Journal in the Jan-Feb 2004 issue, "Asset Allocation With Inflation-Protected Bonds".

I have two questions regarding that article:

- given that over three years have past since you assembled that projections used in the article, and we now have ten years of actual TIPS data to consider, would you advise that I use the data depicted in the graph on page 7 (Figure 1, Overlapping Annual Returns Series....A. Nominal Returns) for my analysis of various multi-asset class portfolios?

- if so, would you mind sending the dataset in a text file or Excel format? (otherwise my plan is to fit gridlines to the graph and try to pick off year-end numbers by eye)

Thanks,



..his reply:

Dear Mr. Xxxx,

yes, you may use the data from the graph, and also you will have to make do with picking-off year-end numbers by eye. The analysis was done by a research assistant and the data are not readily available. Since you are focusing on year-end numbers, reading them off the chart is not a bad idea.

Best wishes,

SP


So here is what I did with the Kothari paper:

I zoomed then copied the nominal returns chart from the PDF,

Pasted it into PaintShop Pro, leveled it slightly, erased the Zero Bonds data, punched up the contrast & darkened the synthetic TIPS lines, then saved as a jpeg.

I then pulled the jpeg into Powerpoint, fitted dashed lines for Years & Returns, then eye-balled the intersections of the Kothari data at year-end for 1954-2001

http://gnobility.com/ER/Kothari_chart_data-rip.ppt

Here's the resultant list of data in Excel:

http://gnobility.com/ER/Kothari_Synthetic_TIPS_data.xls

My inclination is to use the 1972 thru 2001 "simulated" data, and the actual TIPS returns for 2002-2006, because I think the simulated returns might be the better representation, as the earliest TIPS pricing as they were being introduced is probably a poor indicator of their future behavior.

Thoughts/comments?

Cb

User avatar
Barry Barnitz
Wiki Admin
Posts: 3102
Joined: Mon Feb 19, 2007 10:42 pm
Contact:

Re: A simulated historical TIPS data set...

Post by Barry Barnitz » Wed Feb 28, 2007 1:34 pm

Cb wrote:

Here's the resultant list of data in Excel:

http://gnobility.com/ER/Kothari_Synthetic_TIPS_data.xls
Instead of using the arithmetic average of returns you should probably use the compound return (this will be lower than the average because of the volatility drain.)
Image | blb | December Birthday Celebration: Ludwig van Beethoven

User avatar
CyberBob
Posts: 3238
Joined: Tue Feb 20, 2007 2:53 pm

TIPS vs. Nominal Treasuries

Post by CyberBob » Wed Feb 28, 2007 1:45 pm

The first thing I did was slap down 5-year treasury returns right next to your synthetic TIPS returns. Here's what I found:

Code: Select all

January 1, 1953-December 31, 2001 (the period you provided numbers for)

Starting Value=$10,000

                                         TIPS               5-Year Treasury
Ending Value                           $155,979                $227,754
Internal Rate of Return                 5.76%                   6.58%
TIPS don't seem to look that good.
Last edited by CyberBob on Wed Feb 28, 2007 6:12 pm, edited 1 time in total.

BrianTH
Posts: 1340
Joined: Tue Feb 20, 2007 6:10 pm

Post by BrianTH » Wed Feb 28, 2007 2:07 pm

CyberBob,

That certainly looks like a pretty healthy "insurance premium" (as we have been calling the difference in long term expected return between TIPS and comparable nominal treasuries). And should the maturity on your comparable treasuries be higher? I don't know--I'm just asking.

User avatar
CyberBob
Posts: 3238
Joined: Tue Feb 20, 2007 2:53 pm

TIPS vs. Nominal Treasuries

Post by CyberBob » Wed Feb 28, 2007 2:33 pm

BrianTH wrote:And should the maturity on your comparable treasuries be higher?
Okay, here's a chart with the government long-bond added, as well as standard deviation:

Code: Select all

January 1, 1953-December 31, 2001 (the period you provided numbers for)

Starting Value=$10,000

                                  TIPS            5-Year Treasury            Govt Long-Bond
Ending Value                    $155,979             $227,754                    $196,124
Internal Rate of Return         5.76%                  6.58%                       6.26%
Standard Deviation              6.44%                  6.42%                       10.90%
Hello? Yes, I'm calling to cancel my insurance. :wink:
Last edited by CyberBob on Wed Feb 28, 2007 6:14 pm, edited 1 time in total.

BrianTH
Posts: 1340
Joined: Tue Feb 20, 2007 6:10 pm

Post by BrianTH » Wed Feb 28, 2007 2:46 pm

Wait, longer bonds had a LOWER return? That doesn't make much sense (I don't think we have had 50 years of an inverted yield curve).

User avatar
Topic Author
Cb
Posts: 274
Joined: Wed Feb 21, 2007 12:39 am

Re: A simulated historical TIPS data set...

Post by Cb » Wed Feb 28, 2007 3:06 pm

Barry Barnitz wrote:
Cb wrote:

Here's the resultant list of data in Excel:

http://gnobility.com/ER/Kothari_Synthetic_TIPS_data.xls
Instead of using the arithmetic average of returns you should probably use the compound return (this will be lower than the average because of the volatility drain.)
Thanks Barry...I've been saving the TrevH threads from the other board on how to do that.

Cb

User avatar
Random Musings
Posts: 5375
Joined: Thu Feb 22, 2007 4:24 pm
Location: Pennsylvania

TIPS returns

Post by Random Musings » Wed Feb 28, 2007 4:04 pm

Since they are synthetic returns in the study - I assume the study does not take into account the possibilities that investors can "pick" when they want to purchase TIPS (ala Swedroe):

- only purchase TIPS when the fixed income component is high (for example 2.3%, and higher).

- as the fixed TIPS component goes higher - choose longer duration to "lock" in.

I would believe this would reduce the insurance premium.

Let alone, if you follow Larry S. advice and sell TIPS incrementally as the fixed component goes lower (and go into treasuries) - will that also further reduce the net insurance premium or eliminate it altogether?

RM

User avatar
CyberBob
Posts: 3238
Joined: Tue Feb 20, 2007 2:53 pm

Post by CyberBob » Wed Feb 28, 2007 4:15 pm

BrianTH wrote:Wait, longer bonds had a LOWER return? That doesn't make much sense (I don't think we have had 50 years of an inverted yield curve).
I checked 2 data sources for the return numbers, Gummy and TamAsset, and they both agree. According to TamAsset, they got the numbers from Stock, Bonds, Bills, And Inflation, Ibbotson and Sinquefield, Ibbotson Associates, Chicago

Maybe it's because of the volatility of the long-bond returns. Check out this wild ride:

1977 -0.7
1978 -1.2
1979 -1.2
1980 -4.0
1981 1.9
1982 40.4
1983 0.7
1984 15.5
1985 31.0
1986 24.5
1987 -2.7

BrianTH
Posts: 1340
Joined: Tue Feb 20, 2007 6:10 pm

Post by BrianTH » Wed Feb 28, 2007 5:38 pm

Random Musings,

Of course, you would have to compare all that to what would happen if you were using similar strategies with your nominal bonds.

CyberBob,

I guess that makes sense, but for me it least it was unexpected.

User avatar
CyberBob
Posts: 3238
Joined: Tue Feb 20, 2007 2:53 pm

Interesting

Post by CyberBob » Wed Feb 28, 2007 5:43 pm

Hmmm...that's interesting.

When I view the table I made above with Firefox, the columns are neatly lined up. When I view it with Internet Explorer, they look all jumbled. Microsoft is evil! :twisted:

SmallHi
Posts: 1718
Joined: Wed Feb 21, 2007 6:11 pm

TIPS vs Nominal Bonds (1964-2006)

Post by SmallHi » Wed Feb 28, 2007 6:12 pm

Well, as has been mentioned before, TIPS don't appear that appealing over an entire interest rate cycle.

First, looking at the 1964-2006 period (the longest I have for all bond categories) we see the following returns:

1 month t-bills = +5.8%
6 month t-bills = +6.6%
1 year t-notes = +6.7%
2 year t-notes = +7.4%
5 year t-notes = +7.3%
20 year t-bonds = +7.4%
Inflation bonds = +7.0%

(there is an interest rate term premium, but you only need to go out 2-5 years to capture it, and the biggest rewards come from increasing maturities from just 1 month to 6 months. After that, benefits decline dramatically and flatten out beyond 2 years)

Also, listed below is the annual correlation with inflation:

1 month t-bills = .65
6 month t-bills = .55
1 year t-notes = .40
2 year t-notes = .38
5 year t-notes = (.12)
20 year t-bonds = (.29)
Inflation bonds = .05

Finally, since 1973 (starting date of good global equity data), I have listed the correlations of various fixed income securities with STOCKS:

1 month t-bills = (.07)
6 month t-bills = (.05)
1 year t-notes = (.06)
2 year t-notes = .01
5 year t-notes = .07
20 year t-bonds = .13
Inflation bonds = .33

Based on simulated 65% STOCK/35% BOND portfolios rebalanced annually, I could not find any fixed income security that was not preferable to TIPS based on sharpe ratios.

SmallHI

yobria
Posts: 5978
Joined: Mon Feb 19, 2007 11:58 pm
Location: SF CA USA

Re: TIPS returns

Post by yobria » Wed Feb 28, 2007 6:57 pm

Random Musings wrote:Since they are synthetic returns in the study - I assume the study does not take into account the possibilities that investors can "pick" when they want to purchase TIPS (ala Swedroe):
The problem is there's no way to know when to pick. The future could be inflationary, or recessionary. Real rates could go higher or lower.

Nick

User avatar
Robert T
Posts: 2555
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Tips

Post by Robert T » Wed Feb 28, 2007 10:49 pm

cb,

Thanks for the paper and your effort to reproduce the 'synthetic' index.

The characteristics of the series in comparison to 5yr T-Notes seems to vary by time period. For example from 1975 to 2005 (adding Vanguard TIPS returns from 2002-05), the annualized return and standard deviation were 8.33 and 6.93 for 5yr T-Notes and 8.68 and 6.45 for 'synthetic' TIPS i.e. the 'sythestic' TIPS had higher return with lower standard deviation of returns over this period. Adding 50:50 5 yr T-Notes:TIPS to a global small and value tilted portfolio over this period both marginally increased returns and the standard deviation of returns over a portfolio with just 5-yr T-Notes as fixed income.

First post on this new forum - still trying out some of its functions.

Robert

BrianTH
Posts: 1340
Joined: Tue Feb 20, 2007 6:10 pm

Post by BrianTH » Thu Mar 01, 2007 3:23 am

Robert,

That makes sense to me because the mid-late 1970s were probably the best time by far for TIPS relative to nominal bonds.

SmallHi
Posts: 1718
Joined: Wed Feb 21, 2007 6:11 pm

A couple of portfolio simulations...

Post by SmallHi » Thu Mar 01, 2007 10:49 am

Lets assume our base portfolio is 65% global stocks and 35% 5 Year Treasuries. Since 1973, our return/risk profile = 12.8%/11.8

If we substitute TIPS for 5 Year T-Notes, we maintain returns and increase risk: 12.8%/12.4

If we split our allocation evenly between T-Notes and TIPS, we simply split the risk difference: 12.8%/12.0

Finally, if we choose to get our inflation protection from shorter term nominal bonds (2 Year T-Notes), we reduce our portfolio risk slightly: 12.8%/11.6

Now, these are all very long term simulations and probably span one's investing lifetime. My opinion is that it doesn't really matter much. If you are the type that tends to get a bit squirely if your bond portfolio lags the overall bond market by a noticeable amount, then probably better off extending maturities a bit (as 2 Year bonds have had a rough go of it the last 10 - 15 years). If you are primarily concerned with portfolio volatility or have a large % of your portfolio in bonds, you may want to keep it short and sidestep the modestly more volatile longer term T-Notes and TIPS.

Tools for the task? ST Corporate should work for shorter maturities, Int'd Treasury for 5 Year Treasuries, and of course TIPS fund for TIPS.

SmallHI

WorldBeta
Posts: 5
Joined: Thu Sep 27, 2007 4:28 pm

Post by WorldBeta » Sun Jun 01, 2008 4:25 pm

cb - the links are broken, do you have updated ones? thanks

dumbmoney
Posts: 2309
Joined: Sun Mar 16, 2008 8:58 pm

Re: TIPS vs Nominal Bonds (1964-2006)

Post by dumbmoney » Sun Jun 01, 2008 5:24 pm

SmallHi wrote:Also, listed below is the annual correlation with inflation:

1 month t-bills = .65
6 month t-bills = .55
1 year t-notes = .40
2 year t-notes = .38
5 year t-notes = (.12)
20 year t-bonds = (.29)
Inflation bonds = .05
This looks weird. If a consistent definition of "inflation" is used, the synthetic TIPS should have the highest correlation, yes?

User avatar
Topic Author
Cb
Posts: 274
Joined: Wed Feb 21, 2007 12:39 am

Post by Cb » Sun Jun 01, 2008 5:26 pm

WorldBeta wrote:cb - the links are broken, do you have updated ones? thanks
WB,

I uploaded those files again.

I don't have a very high degree of confidence in the synthetic TIPS data I constructed from the Kothari paper. I've since read the Ibottson has constructed a synthetic TIPS dataset, but I haven't seen the dataset itself anywhere on the 'net.

Cb :(

WorldBeta
Posts: 5
Joined: Thu Sep 27, 2007 4:28 pm

Post by WorldBeta » Sun Jun 01, 2008 7:45 pm

i bet they have it in their Encorr software. . .here is a paper from them:

TIPS as an Asset Class
Peng Chen
Ibbotson Associates
225 N. Michigan Ave., Suite 700
Chicago, IL 60601
(312) 616 -1620

User avatar
grabiner
Advisory Board
Posts: 23959
Joined: Tue Feb 20, 2007 11:58 pm
Location: Columbia, MD

Post by grabiner » Sun Jun 01, 2008 9:59 pm

BrianTH wrote:Wait, longer bonds had a LOWER return? That doesn't make much sense (I don't think we have had 50 years of an inverted yield curve).
Long-term bonds can easily have the same return as intermediate-term bonds, because the risk premium is not an issue for many of their purchasers. Pension funds and insurance companies have fixed long-term liabilities, so for them, long-term bonds may be less risky than intermediate-term bonds.

Inversions and steepness in the yield curve are usually noticeable only between short-term and intermediate-term bonds, because the rate on short-term bonds reflects current conditions. The difference between rates on five-year and 20-year bonds doesn't depend much on current conditions, so it doesn't invert or become very steep as often.
Wiki David Grabiner

User avatar
stratton
Posts: 11082
Joined: Sun Mar 04, 2007 5:05 pm
Location: Puget Sound

Post by stratton » Sun Jun 01, 2008 11:38 pm

WorldBeta wrote:i bet they have it in their Encorr software. . .here is a paper from them:

TIPS as an Asset Class
Peng Chen
Ibbotson Associates
225 N. Michigan Ave., Suite 700
Chicago, IL 60601
(312) 616 -1620
Its on the Ibbotson site: TIPS as an Asset Class

Paul

OptionAl
Posts: 386
Joined: Wed Mar 26, 2008 2:05 pm

Post by OptionAl » Mon Jun 02, 2008 1:58 pm

TIPS are fine but for maybe 25% of fixed income. I really don't get it: Why invest in a product that is more volatile and probably going to be outperformed by a similar product and is subject to the whims of the government's inflation computations. I know - diversification and correlations. But when we're told we should invest 50% of fixed income in TIPS, to me it's like hearing we should invest half of our income in insurance.
I like 50%+ IT treasuries/CDs/munis/other, 25% ST (max), 25% TIPS (max). Probably as much real inflation protection as TIPS and some return as well.

adalfu
Posts: 2
Joined: Thu Mar 25, 2010 3:36 pm

Post by adalfu » Tue Apr 27, 2010 8:48 pm

the links in the original post do not work... does anyone have access to these files?

thanks a plenty.

User avatar
Topic Author
Cb
Posts: 274
Joined: Wed Feb 21, 2007 12:39 am

Post by Cb » Tue Apr 27, 2010 9:59 pm

adalfu wrote:the links in the original post do not work... does anyone have access to these files?

thanks a plenty.
I just put them back up.

Cb :|

mlandoni
Posts: 1
Joined: Tue Jan 10, 2012 12:18 pm

Re: A simulated historical TIPS data set...

Post by mlandoni » Tue Jan 10, 2012 1:57 pm

Uhm... the links are gone again. If anyone can post them one more time, I'll figure out something to make them permanent. They are rather interesting data.

thanks!

Post Reply