Black Swan -- Credit Risk

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
Topic Author
SmallHi
Posts: 1718
Joined: Wed Feb 21, 2007 5:11 pm

Black Swan -- Credit Risk

Post by SmallHi »

I took a look at the long term historical behavior of fixed income "credit risk" (as defined by the returns on LT Corporate bonds minus LT Gov't bonds) going back to 1926.

While this may not be a surprise to some, this has been by far the worst 12 month period for corporate bonds relative to gov't bonds in US history dating back to the 20s. With the Lehman Long Term Treasury Index outperforming the Lehman Long Term Corporate Index by over 20% for the 12 months ending 9/30.

In all rolling 12 month periods since 1926 (n=981), there were only seven periods where Credit risk was negative more than 10%. Of those seven periods, the worst was only -14%, with another period of -12%. The other five periods were between -10% and -11%. Put another way, this is about 50% worse than the previously worst ever showing for credit risk! (To put this in perspective, this would be like watching small cap or value stocks underperform large cap or growth stocks by over 60% in one year -- given the worst previous result was only -30% to -40%)

Why might this be surprising to some? Well, of the 5 core multifactor risks (equity, size, price, interest rate, and credit risk), credit had by far the lowest volatility and premium. LT Credit only beat LT Gov't since 1926 by about 0.20% a year annually -- with a standard deviation of about 3.5. That means this event is almost 6 standard deviations away from the mean*!

Certainly there are a number of different take aways from this...happy to hear your thoughts first if you are so inclined.

sh

*I don't feel like doing the math, but I believe a 6 standard deviation event is about 1 in 500M ??
INDUBITABLY
Posts: 312
Joined: Fri Oct 26, 2007 12:01 pm

Post by INDUBITABLY »

Given that you mention Black Swans, the first lesson that comes to mind is that it is better to chase premiums that have an infrequent but large upside than to do the opposite (which would be chasing small credit risk premiums in this case).
"Ah ha! Once again, the conservative, sandwich-heavy portfolio pays off for the hungry investor!" - Dr. Zoidberg
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Post by larryswedroe »

Repeating: Never treat unlikely as impossible or likely as certain

As I have warned, the risk of lower credits is that the correlations tend to rise at the worst time. Check out the long term correlation of the Vanguard high yield fund for example versus the correlation this year!!!
Great example of needing to understand how risks mix and not looking only at correlations over the long term.

Good decisions are result of good process.
Last edited by larryswedroe on Sat Oct 04, 2008 4:44 pm, edited 1 time in total.
INDUBITABLY
Posts: 312
Joined: Fri Oct 26, 2007 12:01 pm

Post by INDUBITABLY »

larryswedroe wrote:Repeating: Never treat unlikely as impossible or likely as certain
Say, Larry, where does that leave people who hold Treasuries as their "safe" money? The widely-held belief is that they cannot default, but given what you just said, it might be worth considering...
"Ah ha! Once again, the conservative, sandwich-heavy portfolio pays off for the hungry investor!" - Dr. Zoidberg
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Post by larryswedroe »

Undubioutedly

First, Treaury will not default on US dollar debt. Countries default on foreign denominated debt. The risk is thus not default but hyperinflation if they print money to repay the debt. Then the risk is revolution (:-((
User avatar
Robert T
Posts: 2803
Joined: Tue Feb 27, 2007 8:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Post by Robert T »

.
Sh,

First on the numbers: I get similar results to you.

Code: Select all

12 months to Oct 3. 2008

US Long-term Credit         -12.96      
US Long-term Treasury*      +11.67
Default premium             -24.63

*78%LB10-20yr:22%LB20yr+Treasury to get the same duration as the US Long-term Credit index of 10.97 yrs.

  • 1926-2005

    Default premium (mean)......0.4%
    Standard deviation..............3.3

    So negative 24.6% seems about 7 to 8 SDs from the mean (1926-2005).

    In general corporate bond returns have a negative skew characteristics. The recent performance seems to re-emphasize this trait.
Second: On Black Swans (events with ‘small probabilities but huge repercussions’)
  • The repercussions of the significant negative default ‘premium’ will likely vary across individuals. For example, for some the largest impact (in the current environment) may show up more significantly as the loss of a job, and/or loss of a home, rather than in portfolio performance – or it could show up in all three dependent on exposure. For others, their total assets may have less exposure to default risk, so for them the repercussions may be small.
On the personal take aways: They are similar to my earlier views expressed in our previous discussions on default/credit risk. There will also likely be revision to the mean at some point. I'm not good at market timing but wish the best to those that try...

Robert
.
peter71
Posts: 3769
Joined: Tue Jul 24, 2007 8:28 pm

Post by peter71 »

Thanks SH, Thanks Robert,

While I'm not generally a fan of people declaring "black swans" it looks like you guys have the sigmas to back it up :D Even if Oct-Oct is a necessary imperfection of the analysis I think you would have had me at 5 . . . in terms of interpretation, I guess my only question would be to what extent this is a phenomenon of commercial credit being bad vs. to what extent this is one of government credit being good . . . that is, are there perhaps other credit crises (in the US and/or internationally) in which the spread has been small because even govt bonds have been no place to hide?

All best,
Pete
RoadTrip!
Posts: 366
Joined: Thu Apr 05, 2007 8:32 am

Folks, perhaps you could speak more plainly??

Post by RoadTrip! »

There are mere mortals who are reading your posts and don't understand what you are talking about (except that it sounds significant). Like me :lol: . Regards, Beth
User avatar
Blue
Posts: 1178
Joined: Sat Jul 12, 2008 10:18 pm

Post by Blue »

Anybody up for the arbitrage of selling treasuries and buying corporate bonds? If I recall correctly, this is the same strategy LTCM used?
peter71
Posts: 3769
Joined: Tue Jul 24, 2007 8:28 pm

Post by peter71 »

Hi Beth,

To translate, it's almost never been the case that GOVERNMENT bonds have done so much better than CORPORATE bonds . . . setting the statistical stuff aside (which just measures how unusual the difference in their relative performance actually is), I'm wondering to what extent that's because corporate bonds have done UNUSUALLY BADLY or, alternatively, because government bonds have done UNUSUALLY WELL.

All best,
Pete
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Post by larryswedroe »

Robet T

Your point is the one I have been making---you MUST consider how your labor capital correlates with the risks of stocks. Unfortunately it is my experience that even most professional advisors don't ask that question thus fail to take it into account.
User avatar
Random Musings
Posts: 6756
Joined: Thu Feb 22, 2007 3:24 pm
Location: Pennsylvania

Post by Random Musings »

Larry S wrote:
First, Treaury will not default on US dollar debt. Countries default on foreign denominated debt. The risk is thus not default but hyperinflation if they print money to repay the debt.
Perhaps diversification of ones cash/bonds in other currencies (yen, swiss franc) or foreign bonds could be in order if the smell of hyperinflation begins at home.....

And if that's the case, it still results in deflation at the end. To that end, wouldn't debt destruction and the resulting deflation be better in the long-run for the U.S versus the hyperinflation/deflation scenario?

RM
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Post by larryswedroe »

RM
hyperinflation is the worse alternative----except for those that are net debtors as their debts get wiped out.

But deflation wipes out debts in another way--bankruptcy.

Neither is good.
Sidney
Posts: 6784
Joined: Thu Mar 08, 2007 5:06 pm

Post by Sidney »

How do the numbers of this analysis change if you look at the short end of investment grade instead of the long end? Does it change materially?
I always wanted to be a procrastinator.
Valuethinker
Posts: 48944
Joined: Fri May 11, 2007 11:07 am

Post by Valuethinker »

larryswedroe wrote:Undubioutedly

First, Treaury will not default on US dollar debt. Countries default on foreign denominated debt. The risk is thus not default but hyperinflation if they print money to repay the debt. Then the risk is revolution (:-((
It wouldn't be revolution. This is the US with its own history-- there's no strong socialist current in American history.

It would be either military rule, or an Adolph Hitler (or Juan Peron) type populist leader. Lots of people strutting around in uniforms.
Valuethinker
Posts: 48944
Joined: Fri May 11, 2007 11:07 am

Re: Black Swan -- Credit Risk

Post by Valuethinker »

SmallHi wrote:I
sh

*I don't feel like doing the math, but I believe a 6 standard deviation event is about 1 in 500M ??
Benoit Mandelbrot "The Misbehaviour of Markets" also George Soros "The Alchemy of Finance"

What these events mean is our theory of the distribution of returns in finance is incorrect (likely in particular the notion that we are engaging in independent sampling of any given sub-period against another) NOT that this is a 1 in 500m event.

What Mandelbrot points out is that normal distribution is assumed for theoretical tractability, but most financial markets do not have outcomes which are normally distributed.
User avatar
Robert T
Posts: 2803
Joined: Tue Feb 27, 2007 8:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Re: Black Swan -- Credit Risk

Post by Robert T »

Valuethinker wrote: What these events mean is our theory of the distribution of returns in finance is incorrect (likely in particular the notion that we are engaging in independent sampling of any given sub-period against another) NOT that this is a 1 in 500m event.

What Mandelbrot points out is that normal distribution is assumed for theoretical tractability, but most financial markets do not have outcomes which are normally distributed.
Mandelbrot and Taleb suggest using the fractile method (rather than the Gaussian ‘normal’ distribution) when looking at financial markets.

My earlier comparison shows more frequency of extreme events using the fractile vs. Gaussian method. Interestingly a portfolio with an expected return of 7% and expected SD of 12 under the fratile method should expect a greater than 30% decline every 96 yrs (its seem about that since the great depression), under the normal distribution, the 30% decline is expected every 720 years…

Robert
.
Topic Author
SmallHi
Posts: 1718
Joined: Wed Feb 21, 2007 5:11 pm

Post by SmallHi »

Thanks! Some interesting thoughts, for sure. A few random ideas that come to mind:

a) this isn't exactly the "normal" course of events in terms of fixed income behavior in the face of crisis/equity declines. In 1987 (an equity black swan), for example, for the 2 months ended in November, the annualized credit premium was -1.7% (the annualized return on the S&P 500 was -86%). Of course, we don't have a very good definition of what "normal" is.

b) many studies have indicated that credit risk is most lucrative and consistent on the short end of the curve. This period indicates somehow that this doesn't necessarily come at the cost of higher risk (relative to long dated credit risk). Here are the 1YR Credit premiums per maturity:

Code: Select all

Long = (20.3%)
Medium = (16.24%)
Short = (8.41)
UltraShort = (5.83)

Long/Medium/Short = Vanguard specified Indexes
Ultra Short = Lehman 1-3YR Treasury/Credit
I don't have the statistics on short dated credit risk, but the deficit is 75% less on the short end than on the long end -- not out of question or completely unexpected. I doubt anyone who favors short term bonds is losing much sleep over a one year underperformance of about 6%.

c) as was mentioned above, this credit event isn't due to the abnormally good returns of gov't bonds, its due to the abnormally bad returns of corporate bonds (primarily on the longer end). But this is part of a somewhat normally expected distribution on their part. Since 1926, they have compounded at +5.8% a year, with a deviation of 8.5%. So once every 100 years, you'd expect a 20% one year decline, and a 31% gain. The reality? The worst 1YR return for LT Corps was -18%. The best 1YR return for LT Corps was +50%. The distribution of historical 1YR LT Credit returns is positively skewed

d) there is still obvious differentiation between Investment grade credit risk and junk bond credit risk. The Junk - Credit premium in the last 12 months (-5.3%) is not as large as the Credit - Treasury premium (-16.2%), and obviously wider than expected. But there is still some acknowledgement from market participants that not all credit risk is created equal.

e) even today, there is an historical credit premium on medium/shorter duration indexes (albiet minor):

Code: Select all

1973-2008

Lehman Credit Intermediate = +8.44%
Lehman Treasury Intermediate = +8.2%
Obviously that is not statistically significant, but given the time frame (bookended by the two worst market/economic enviroments since the great depression), it doesn't appear like high end credit risk is a huge mistake (of course some would argue its not worth it)

There is, however, no Junk premium that I could find over a slightly shorter period:

Code: Select all

1978-2008

Vanguard HY Corporate fund = +8.45%
Lehman Credit Intermediate = +8.44%
f) there is no indication that index funds/structured funds are reliably unable to track their indexes even in periods of bond market duress (the last 12 months ended 9/30):

Code: Select all

Vanguard LT Credit fund minus LT Index = +4.6%
Vanguard INT Credit fund minus INT Index = +2.4%
Vanguard ST Credit fund minus ST Index = +0.4%
iShares 1-3YR Credit fund minus 1-3YR Index = (0.1%)
In all cases but the shortest end, investors haven't actually earned the "black swan", as their has been positive tracking error on the credit side and negative tracking error (not shown, but all treasury portfolios have slightly underperformed their indexes) on the treasury side.

g) you can take your treasury regret as far as you'd like, but I'd be hard pressed to say now is the right time to endorse your newfound love of treasuries. Like Robert, I am a horrible market timer. But Vanguard ST Investment Grade has a published YTM of 4.9% vs. 2.1% for Vanguard ST Treasury. Yes, the equity risk premium has widened as well, but if you are going to move to treasuries today, you best lower your fixed income allocation at the same time, or you'll likely be wipsawed both ways.

h) <thinking out load> are we witnessing a market event where the historical nature of the fixed income markets are changing?

From 1926 through 1981, there was no real return for LT Gov't bond holders -- despite considerable short term volatility and (obvious) inflation risk. Yields were driven up to the point of a considerable long term future interest rate premium in the last 25 years (an annualized TERM premium of 5.5% a year <which is equity premium like in its payoff> and a 7.5% real return on LT Gov't bonds)

Are we witnessing something similiar in the credit markets? Certainly no one would argue that credit risk (certainly long dated credit risk) has been fairly priced in the last 80 years. What would it take to reasonably assume the market was making an effort at moving towards equilibrium in terms of how its pricing this risk? Obviously a huge one time widening of the spread between Treasury/Credit yields (and the painful corresponding price movements for credit investors).

anyway, thanks for the thoughts. feel free to agree or disagree.

sh

PS -- LT treasury bonds may not mix well in a traditional mean variance framework, and may require being paired with copious amounts of TIPS to manage their inflation risk, but you have to be blind to not see what Swensen was talking about when he discussed (as Robert T has so often pointed out correctly) the superior diversification <read: downside risk protection> of long term US Treasury bonds
Valuethinker
Posts: 48944
Joined: Fri May 11, 2007 11:07 am

Re: Black Swan -- Credit Risk

Post by Valuethinker »

Robert T wrote:
Valuethinker wrote: What these events mean is our theory of the distribution of returns in finance is incorrect (likely in particular the notion that we are engaging in independent sampling of any given sub-period against another) NOT that this is a 1 in 500m event.

What Mandelbrot points out is that normal distribution is assumed for theoretical tractability, but most financial markets do not have outcomes which are normally distributed.
Mandelbrot and Taleb suggest using the fractile method (rather than the Gaussian ‘normal’ distribution) when looking at financial markets.

My earlier comparison shows more frequency of extreme events using the fractile vs. Gaussian method. Interestingly a portfolio with an expected return of 7% and expected SD of 12 under the fratile method should expect a greater than 30% decline every 96 yrs (its seem about that since the great depression), under the normal distribution, the 30% decline is expected every 720 years…

Robert
.
Are you not assuming that each sample point is independent, identically distributed?

Because they are not. Trading days show correlation. And a drop, by precipitating a liquidation, can causally cause another drop (bad English).
User avatar
Robert T
Posts: 2803
Joined: Tue Feb 27, 2007 8:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Post by Robert T »

.
Sh,

Few additional thoughts...
I doubt anyone who favors short term bonds is losing much sleep over a one year underperformance of about 6%.
Maybe, maybe not. I recall several recent threads on the Vanguard ST investment grade fund highlighting significant concern…
The worst 1YR return for LT Corps was -18%. The best 1YR return for LT Corps was +50%. The distribution of historical 1YR LT Credit returns is positively skewed
If we decompose the source of the skewness (using Ibbotson data)

LT corporate returns are simply the annual returns of the risk free rate (Rf) + 1*default premium + 1*term premium (by identity definition)

Code: Select all

1927-2005

Skewness (+ve indicates positively skewed, -ve indicates negative skewness)

Rf          = +0.97
Default     = -0.20
Term        = +0.56
Sum         = +1.33

While skewness is not additive it coincidentally comes fairly close over the 1927-2005 time period. The actual skewness of the annual returns of LT corporates was +1.39 from 1927-2005. So default risk has historically been negatively skewed (more so if recent returns are added). Not a characteristic I particularly like in fixed income (particularly when it shows up at the same time as equity market (and other personal asset) declines.
In all cases but the shortest end, investors haven't actually earned the "black swan", as their has been positive tracking error on the credit side and negative tracking error (not shown, but all treasury portfolios have slightly underperformed their indexes) on the treasury side.
But not enough to offset the significant difference (they still earned a significant portion of the negative default premium).

Code: Select all

                       Short     Intermediate    Long  

Vanguard Treasury      +4.46       +6.36        +6.49
Vanguard Corporate     -2.21       -5.92        -7.39
Default premium        -6.67      -12.55       -13.88 

Finally, I agree its important to stick to long-term asset allocation targets. Whatever investors decide these targets to be, they need to be comfortable with them (the risk they are exposed to) to stay with it for the long-term.

Just on Swensen’s view on fixed income – this is from the 2005 Yale Endowment Report.
Sensible investors focus on the diversifying characteristics of long-term government bonds, holding only the amount necessary to protect portfolios against financial trauma. If portfolios include the minimum allocation necessary to provide insurance against catastrophe, investors free up assets to diversify into alternative asset classes, achieving volatility reduction without sacrificing return. A low allocation to high-quality fixed income reduces the costs associated with holding bonds during normal circumstances and periods of unanticipated inflation, the environments in which fixed income positions tend to impair portfolio performance. Tailoring the bond portfolio to emphasize fixed income’s essential diversifying characteristics increases expected benefits in time of crisis, while reducing the long-term costs of holding bonds.
By long-term, I think he means/includes intermediate term, as the fixed income benchmark for the Yale Endowment is the US Treasury Index. Nevertheless, his message seems to be keep fixed income in highest quality – take risk with equities not fixed income. Not too dissimilar from Fama Jr. IMO.


_____________________________________


Valuethinker,
Are you not assuming that each sample point is independent, identically distributed?

Because they are not. Trading days show correlation. And a drop, by precipitating a liquidation, can causally cause another drop (bad English).
In my earlier example, my understanding is it does assume each annual observation is independent. Not perfect but probably better than the Gaussian normal distribution. Also assumes no revision to the mean relations…a point made by Bernstein in MC simulation results of dispersion of likely returns.

I also found Gratham’s earlier views on this interesting here are some extracts.

Robert
.
User avatar
BlueEars
Posts: 3968
Joined: Fri Mar 09, 2007 11:15 pm
Location: West Coast

Post by BlueEars »

Robert T wrote:...Just on Swensen’s view on fixed income – this is from the 2005 Yale Endowment Report.
Sensible investors focus on the diversifying characteristics of long-term government bonds, holding only the amount necessary to protect portfolios against financial trauma. If portfolios include the minimum allocation necessary to provide insurance against catastrophe, investors free up assets to diversify into alternative asset classes, achieving volatility reduction without sacrificing return. A low allocation to high-quality fixed income reduces the costs associated with holding bonds during normal circumstances and periods of unanticipated inflation, the environments in which fixed income positions tend to impair portfolio performance. Tailoring the bond portfolio to emphasize fixed income’s essential diversifying characteristics increases expected benefits in time of crisis, while reducing the long-term costs of holding bonds.
By long-term, I think he means/includes intermediate term, as the fixed income benchmark for the Yale Endowment is the US Treasury Index. Nevertheless, his message seems to be keep fixed income in highest quality – take risk with equities not fixed income. Not too dissimilar from Fama Jr. IMO.
Robert, any thoughts on why Swensen is ignoring TIPS? Does he feel that TIPS are not as good a deflation fighter as LT Treasury? We have recently seen TIPS yields going up after declining dramatically along with Treasuries. Also, what is the duration roughly of US Treasury Index?
User avatar
Robert T
Posts: 2803
Joined: Tue Feb 27, 2007 8:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Post by Robert T »

.
Les,
any thoughts on why Swensen is ignoring TIPS?
For institutional portfolios, may be he feels other assets do a good enough job...i.e real estate, oil and gas, and timberland. From the same 2005 Yale Endowment Report.
  • "Real estate, oil and gas, and timberland share common characteristics: sensitivity to inflationary forces, high and visible current cash flow…[These] Real assets investments provide attractive return prospects, excellent portfolio diversification, and a hedge against unanticipated inflation."
For individual portfolios, he seems to like TIPS. From unconventional success…
  • On TIPS: "The combination of the default-free character of full-in-faith obligations of the US government and the mathematically certain protection against inflation provides investors with a powerful portfolio tool.”

    "Investors who desire more certain protection from inflation increase the US Treasury TIPS allocation. Investors who require greater protection against financial crisis expand US Treasury bond exposure."
Does he feel that TIPS are not as good a deflation fighter as LT Treasury?
Yes, I think so. From Unconventional Success...
  • "Unexpected deflation helps regular bonds by increasing the purchasing power of the fixed income stream of payments. In contrast, unexpected deflation reduces the stream of periodic interest payments to holders of TIPS, even though deflation fails to reduce the final principal payment."
Also, what is the duration roughly of US Treasury Index?
Hope this is helpful.

Robert
Last edited by Robert T on Sun Oct 05, 2008 7:21 pm, edited 1 time in total.
User avatar
BlueEars
Posts: 3968
Joined: Fri Mar 09, 2007 11:15 pm
Location: West Coast

Post by BlueEars »

Robert T wrote:...From Unconventional Success...

"Unexpected deflation helps regular bonds by increasing the purchasing power of the fixed income stream of payments. In contrast, unexpected deflation reduces the stream of periodic interest payments to holders of TIPS, even though deflation fails to reduce the final principal payment."
Robert, thanks. I guess the above quote makes plenty of sense. Perhaps the rising TIPS yields now are telling us something about "unexpected deflation". Whenever I try to explain the rising real rates we've had recently I just get confused.
detifoss
Posts: 556
Joined: Sat Oct 13, 2007 10:38 pm

Post by detifoss »

thanks (especially to SmallHi and Robert T)

this thread has made great reading and outstanding education

I think the term Black Swan is overused, however, the analysis you both present here is rather compelling that at least part of this financial crisis (the credit portion) is truly historic (and very black or is it red?)

thanks again
Post Reply