The Great Debate!

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Topic Author
Rick Ferri
Posts: 9209
Joined: Mon Feb 26, 2007 11:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

The Great Debate!

Post by Rick Ferri »

Lincoln-Douglas. Nixon-Kennedy... And Now

The Great Commodities Debate!

Coming to your computer screen starting February 12.
User avatar
SoonerSunDevil
Posts: 2000
Joined: Mon Feb 19, 2007 10:32 pm
Location: The desert

Post by SoonerSunDevil »

Oh, no... :lol:

Rick, I cannot believe you and Larry are only going to be discussing commodities! Why not stage a three-part debate? Begin with commodities, then move to high-yield, and then have a finale with "mirror image" portfolio constructions. Personally, I would recommend you and Larry find agents and battle it out on Pay-Per-View cage match style, but I guess the internet will have to do.

Not that either you or Larry care, but I agree with Larry on the role of bonds in a portfolio, you on commodities, and have not formed an opinion on "mirror image" locations of assets for those who are emotionally unable to cope with multiple accounts forming the portfolio.

Most Admirably,

John
User avatar
Taylor Larimore
Advisory Board
Posts: 29952
Joined: Tue Feb 27, 2007 8:09 pm
Location: Miami FL

Commodities ?

Post by Taylor Larimore »

Hi Rick:

Thanks for the link. We will be looking forward to you and Larry sharing your insights about using commodities in our portfolios.

Best wishes.
Taylor
User avatar
tetractys
Posts: 4831
Joined: Sat Mar 17, 2007 3:30 pm
Location: Along the Salish Sea

Post by tetractys »

A question for both Rick and Larry:

Does portfolio value make any difference? In other words, would the pros and/or cons of commodities in a portfolio change with larger or smaller portfolios. And by portfolio here I mean long-term buy-and-hold.

This broad question has been on my mind for awhile. If it comes up in your debate, cool.

My gut says yes. For larger portfolios commodities could be of benefit; but with smaller portfolios, commodities would actually be a detriment. (I have no idea where the line between larger and smaller would be.)

Best wishes, Tet

& BTW, thanks for the announcement.
User avatar
Topic Author
Rick Ferri
Posts: 9209
Joined: Mon Feb 26, 2007 11:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

Post by Rick Ferri »

Does portfolio value make any difference? In other words, would the pros and/or cons of commodities in a portfolio change with larger or smaller portfolios. And by portfolio here I mean long-term buy-and-hold.
Heck no! Stick with stocks and bonds. The stuff that gives you 'real' returns.
User avatar
ddb
Posts: 5511
Joined: Mon Feb 26, 2007 12:37 pm
Location: American Gardens Building, West 81st St.

Post by ddb »

I can't imagine how this upcoming debate could be more comprehensive than the volumes of posts between you two on this subject over the years!
"We have to encourage a return to traditional moral values. Most importantly, we have to promote general social concern, and less materialism in young people." - PB
User avatar
Boris
Posts: 552
Joined: Tue Feb 27, 2007 11:10 am
Location: CT

Post by Boris »

Hmm.. I'm looking forward to this, however... where's the debate?

"...commodities fail the expected real return test. That's why I don't include them in portfolios. The lower expected returns of commodity funds is not a fair trade for the proven real returns of assets such as stocks."
- Rick Ferri

"Futures, or more specifically CCFs, are one of those rare asset classes that have negative correlation to both stocks and bonds...With negative correlation we actually like higher volatility because, with rebalancing, the diversification return increases."
- Larry Swedroe

Rick is quoted as talking about owning direct commodities whereas Larry talks about futures (derivatives that can almost be thought of as "insurance" to the producers). Isn't that like talking about investing in a building versus an insurance product which insures that building?

Am I not understanding something?

Boris
Short term moves in the market are like "a tale Told by an idiot, full of sound and fury, Signifying nothing." | - John C. Bogle quoting Shakespeare
User avatar
stratton
Posts: 11083
Joined: Sun Mar 04, 2007 5:05 pm
Location: Puget Sound

Post by stratton »

As a sideshow we could get a Rock'em Sock'em Robots game off eBay and have Larry and Rick go at it in that "forum." :P

Paul
User avatar
Topic Author
Rick Ferri
Posts: 9209
Joined: Mon Feb 26, 2007 11:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

Post by Rick Ferri »

Am I not understanding something?
Commodities, futures, same-same in the long-term.

There is no free-money tree for futures buyers. Futures are a side bet on future spot prices. For every winner there is a loser. While there has been a recent period of excess return to the buyers, in the very long run the expected return from commodities and futures is the same, i.e. the inflation rate minus costs.

Sorry, can't say anymore.

Rick Ferri
User avatar
Boris
Posts: 552
Joined: Tue Feb 27, 2007 11:10 am
Location: CT

Post by Boris »

Rick Ferri wrote:
Am I not understanding something?
Commodities, futures, same-same in the long-term.

There is no free-money tree for futures buyers. Futures are a side bet on future spot prices. For every winner there is a loser. While there has been a recent period of excess return to the buyers, in the very long run the expected return from commodities and futures is the same, i.e. the inflation rate minus costs.

Sorry, can't say anymore.

Rick Ferri
Fair enough, I'll be tuning in!!

Thanks,

Boris
Short term moves in the market are like "a tale Told by an idiot, full of sound and fury, Signifying nothing." | - John C. Bogle quoting Shakespeare
User avatar
tetractys
Posts: 4831
Joined: Sat Mar 17, 2007 3:30 pm
Location: Along the Salish Sea

Post by tetractys »

stratton wrote:As a sideshow we could get a Rock'em Sock'em Robots game off eBay and have Larry and Rick go at it in that "forum." :P

Paul
Or this might be one of those occasions where a shot after each comment is in order. Last man standing wins. :lol:

Or put a quarter in the bottom of a cup.... :lol:

Peace, Tet
RESISTANCE IS FRUITFUL
peter71
Posts: 3768
Joined: Tue Jul 24, 2007 8:28 pm

Post by peter71 »

hmm, /i'm/ definitely interested but it seems to me like a pretty technical debate in that much of what the two of you seem to be arguing about is whether correlations between commodities and stocks & bonds remain relatively stationary (consistent) across various "relevant" time periods (e.g. do they become more positive in bear markets when you especially want them negative and more negative in bull markets when you especially want them positive, and, finally, what about the overall period(s) of investors' time horizons). . . data on these issues could definitely be complied, but as pearson-r correlation coefficients are considerably less intuitive than, e.g., OLS regression coefficients (compare the wikipedia entries on correlation and regression analysis) i think it'd be particularly helpful if one or both of you could put things in historical "portfolio impact" terms by period.

all best,
pete
User avatar
AzRunner
Posts: 999
Joined: Mon Feb 19, 2007 6:18 pm
Location: Phoenix

Post by AzRunner »

OUJohnNasr wrote:Not that either you or Larry care, but I agree with Larry on the role of bonds in a portfolio, you on commodities, and have not formed an opinion on "mirror image" locations of assets for those who are emotionally unable to cope with multiple accounts forming the portfolio.

Most Admirably,

John
I'll weigh in with my votes as well:
Commodities: Rick - I buy the argument that inflation rate returns do not justify commodities. Just buy TIPS.
Bonds in a Portfolio: I go with Larry. I can see that bonds are no different from stocks in terms of efficiency and modern portfolio theory but I just want to have a safe haven for my fixed income.
Mirror Image: I'm with Larry on this one. I minimize the taxes I owe by having bonds in tax deferred and efficient equity in taxable. I rebalance based on selling things since we need a portion of the portfolio for living expenses.

Norm
Sam I Am
Posts: 2062
Joined: Tue Feb 20, 2007 7:58 pm

Sounds like fun..... I think!

Post by Sam I Am »

Message deleted.
Last edited by Sam I Am on Sat Oct 05, 2013 1:41 pm, edited 1 time in total.
Eric White
Posts: 54
Joined: Fri May 18, 2007 10:09 am

LET THE GAMES BEGIN!

Post by Eric White »

I'm definitely planning on polling the forum after the debate.

Here are my votes before the debate (I still need to do more thread reviews too):
Commodity Inclusion in Asset Allocation: NO (Rick 1, Larry 0)
Bond Asset Subclass Diversification: YES (Rick 1, Larry 0)
Portfolio Mirroring: YES (Rick 1, Larry 0)

At round 0, my grand total is:
Rick 3, Larry 0

Is there a difference in positions on Larry's tactical bond allocation as well? I'm not as well up to speed on that subject.

LET THE GAMES BEGIN!

Cheers,
Eric
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Post by larryswedroe »

tet
The size of the portfolio really doesnt matter-what matters is risk aversion

Boris-What Rick continually ignores, despite all the evidence, is that the high volatility of the individual components of a commodity index like the DJ AIG and there low correlation provides the same type of diversification return (only with far more certainty-not my opinion but the results from several academic papers) that say EM and int'l small provide which results in significant value added to the portfolio--way beyond the return of the asset class itself. That has historically been extremely high. And there is no reason to think that won't be repeated. Even if it was much smaller there would still be value.

This is why there is a big difference between the return on commodities themselves which individually or weighted average of the individual commodities in real terms is expected at zero but the return on CCF is not. That is point Rick ignores--and that is despite the very paper he cites says is the most reliable part of the return--on order of 3-5% p.a.

And the same is true for the issue related to commodities and how they mix with equities and bonds. Rick has continuously ignored how CCF mix with bonds--and how adding CCF can allow an investor to extend maturities, picking up return, yet hedging the inflation risk of longer bonds. That is another example of why one should consider an asset in how it impacts total portfolio and not in isolation.

The second error is to me shocking that any professional would say it--that one should look at the return of the asset class and not the impact the asset class has on the risk and return of the entire portfolio. There is absolutely no basis for that statement. It is simply wrong. And portfolio theory tells us that negatively correlating assets should have less than riskless returns due to their insurance nature. That point alone shows how incorrect Rick's statements are. As I have shown, even during a period when commodities significantly underperformed equities they added significant value to portfolios.

There may be good reasons for not including CCF but the ones Rick site don't hold up to scrutiny,

For others
The issue of negative correlation and why one should be able to rely on it for CCF is simple--stocks and bonds are negatively correlated to inflation and CCF are positively correlated. Also stocks are negatively correlated with event risks while CCF often is positively correlated (though not always--which is why CCF is better hedge even against bond risk than equity risk)

So you judge for yourself.
User avatar
dave.d
Posts: 936
Joined: Mon Mar 19, 2007 10:30 pm
Location: Richmond, VA

Post by dave.d »

Well, if it makes Larry feel any better after Eric's post, I favor Larry on all 3 issues, although...

- My allocation to commodities is small. I share Bill Bernstein's concern that they're a little too hip and popular right now, a little too-cool-for-school, the asset-class-du-jour that's about to tank. I'll probably up my allocation when they do.

- On bonds I'll do investment grade, just not junk... but I'm beginning to be tempted by the spread. And I certainly can't abide the idea of holding treasuries at current yields.

- I'm close to neutral on mirroring. I view allocation across taxable and non-taxable accounts as just a very complex math problem. To do it and rebalance it properly is probably beyond a lot of investors who wouldn't follow a forum like this. (I have a spreadsheet.) Mirroring may be sub-optimal but more practical, if it permits people to hold and maintain otherwise proper allocations. Plus I see some advantage in having a small portion of my stocks in tax-protected so I can rebalance tax-free the next time the market makes an upward run. I'm down to 18% of my stocks in tax-protected, thus a lower proportion of tax-protected than of taxable. If I've gauged it right I'll exhaust that just as I retire in a few years.

But then, I like Rick's books better. All in all, it's Spy vs. Spy!
Value-based allocation.
timid investor
Posts: 598
Joined: Thu Sep 27, 2007 9:05 pm
Location: alabama

Post by timid investor »

Well let's put it this way.
I'm 100% equities, emerging markets to be precise.
When I want to hedge against inflation, larry's arguments for CCF's make much more sense than Rick's caveats in preserving portfolio real returns.


I'm considering an 80/20 em /ccf's
or 50/30/20 emerging/frontier/ccf when available
peter71
Posts: 3768
Joined: Tue Jul 24, 2007 8:28 pm

Post by peter71 »

larryswedroe wrote:
For others
The issue of negative correlation and why one should be able to rely on it for CCF is simple--stocks and bonds are negatively correlated to inflation and CCF are positively correlated. Also stocks are negatively correlated with event risks while CCF often is positively correlated (though not always--which is why CCF is better hedge even against bond risk than equity risk)

So you judge for yourself.
Hi Larry,

I think event shocks in particular would be difficult (though not impossible) to quantify, but insofar as you're suggesting that a portfolio without CCF might outperform in nominal but not in real terms that's an interesting argument, but are you then conceding that a non-CCF portfolio would "win" given a quick back of the envelope comparison of nominal return and nominal SD? No need to have the debate now, of course, and having not seen much systematic data I'm entirely undecided myself :D

All Best,
Pete
User avatar
Topic Author
Rick Ferri
Posts: 9209
Joined: Mon Feb 26, 2007 11:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

Post by Rick Ferri »

What Rick continually ignores, despite all the evidence, is that the high volatility of the individual components of a commodity index like the DJ AIG and there low correlation provides the same type of diversification return that say EM and int'l small provide which results in significant value added to the portfolio--
Very wishful thinking, IMO. Only in academia. Very unreliable way to invest. Giving up the good return of stocks and bonds for the hope of returns from a zero real returning asset class over the long run, which the exception of potential excessive volatility and maybe low correlations in the future perhaps giving you some small rebalancing benefit. Bad idea.

Larry's mistake is that he considers low correlation as the best reason to invest in an asset class. He forgets that you have to give up the great long-term returns from stock to invest in the lousy long-term returns from commodities. That is not the way I believe you should invest. I believe ALL your investments should earn a real return on their own. If they also assist in adding return to the portfolio because of low correlation, then that is a BENEFIT, not the reason to invest in it.

Commodities are good for a speculative play. Put your money down and roll the dice. If you hit the lucky streak, you win. If you don't, it is dead money for a long, long time. BUT, even if you lose, the good news according to Larry, is that the dead money has low correlation to stocks and bonds. So, just like burying money in glass jars in the backyard, you should invest in them for the low correlation with stocks. No, no, no.

Rick Ferri

BTW, commodity funds are about 10 times more expensive than stock index funds. That is a little important, don't you think? I sure that will be brought up in the debate.
peter71
Posts: 3768
Joined: Tue Jul 24, 2007 8:28 pm

Post by peter71 »

Rick Ferri wrote:
What Rick continually ignores, despite all the evidence, is that the high volatility of the individual components of a commodity index like the DJ AIG and there low correlation provides the same type of diversification return that say EM and int'l small provide which results in significant value added to the portfolio--
Very wishful thinking, IMO. Only in academia. Very unreliable way to invest.
Rick,

I agree with your final point about current valuations but as I'm not sure other Bogleheads will I thought you had a more nuanced critique of relying on correlations data than "only in academia" . . . I don't presume you're saying that Sharpe/Markowitz's data is from some parallel universe, are you? Alternatively, if it's that you don't like SD as MPT's indicator of risk that's fair enough, but that takes one down what to me is an even more technical road, and I guess I'll stand up for academia in saying being able to travel down those roads might not always be of trivial importance . . .

All best,
Pete
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Post by larryswedroe »

Rick continues to mistate what I say and what I do. Here is another example

"Larry's mistake is that he considers low correlation as the best reason to invest in an asset class. He forgets that you have to give up the great long-term returns from stock to invest in the lousy long-term returns from commodities."

Now I challenge Rick to show me where I made that first statement. Or if he cannot find it I challenge him to apologize for the mistatement

I never said any such thing. What I did say is that it is not sufficient to consider only return and SD in isolation but to consider how the addition of an asset class impacts the return of the portfolio. That is nothing like what Rickclaims I said

And what I state is exactly what any academic with any training would state beginning with Markowitz who got Nobel Prize for that observation. Seems Ferri has outlawed the most basic precept of MPT.
User avatar
Jazztonight
Posts: 990
Joined: Wed Feb 28, 2007 12:21 am
Location: Lake Merritt

Post by Jazztonight »

Two questions:

1. Why do I feel like I'm watching a program on Fox TV?

2. Does spelling count?

This debate is actually quite entertaining AND informative. I have the greatest respect for both Larry and Rick, enjoy (and profit from) reading their books, and thank them for being part of this Forum.
"What does not destroy me, makes me stronger." Nietzsche
Eric White
Posts: 54
Joined: Fri May 18, 2007 10:09 am

Initial review in anticipation of the big match

Post by Eric White »

Alright folks. Just like in all the other threads, I can only read so many opinions without data before I get a little bug-eyed. :shock: Last time I looked I thought this was the VANGUARD DIEHARDS forum. WOOT! 8)

Here is the current Simba/TrevH asset allocation backtest file with candidate alternative asset subclass portfolio results:
http://groups.google.com/group/vanguard ... %2C+v4.xls

ASSUMPTIONS:
- Slice & Dice
- Value Tilt
- Alternative Asset Subclass = 10% of overall portfolio (excluding REITs)

ALTERNATIVE ASSET SUBCLASS PORTFOLIO CANDIDATES:
A = 10% Commodities (PCRIX)
B = 10% Energy Fund (VGENX)
C = 10% Health Care Fund (VGHCX)
D = 10% Precious Metals Fund (VGPMX)
E = 10% Emerging Mkts (VEIEX) ***incremental to previous input***
F = 2.5% Commodities / 2.5% Energy / 2.5% Health Care / 2.5% Precious Metals
G = 3.3% Energy / 3.3% Health Care / 3.3% Precious Metals
H = 2% Commodities / 2% Energy / 2% Health Care / 2% Precious Metals / 2% incremental to Emerging

RESULTS (CAGR Nominal / CAGR Real / Max Drawdown Rebalanced / Sortino Nominal / Sortino Nominal):
A - 14.4 / 11.0 / -11.2% / 2.3 / 1.2
B - 14.4 / 11.0 / -15.0% / 1.7 / 1.0
C - 15.0 / 11.6 / -13.5% / 2.0 / 1.2
D - 14.5 / 11.1 / -15.7% / 1.9 / 1.1
E - 15.3 / 11.8 / -16.0% / 2.1 / 1.2
F - 14.6 / 11.2 / -13.5% / 1.9 / 1.1
G - 14.6 / 11.2 / -14.3% / 1.9 / 1.1
H - 14.7 / 11.3 / -13.8% / 2.0 / 1.1

The only two portfolios to have BOTH enhanced return and low risk relative to median results are:
C
H

I personally would be a little hesitant to place all my eggs in one basket (C=Health Care), thus I would probably prefer H (even spreading of allocation across alternative asset subclasses).


DIRECT OBSERVATIONS:
1.) Rick Ferri seems correct in his claim that Commodities, relative to other alternative asset subclasses, hurt your nominal and real returns. Commodities provided the worst CAGR Nominal and Real rates of Portfolios A through H.
2.) Larry Swedroe seems incorrect in his claim that the low correlation of Commodities as an alternative asset subclass within a portfolio overcomes this impact to nominal and real returns.
3.) Larry Swedroe seems correct in his claim that Commodities are the best asset subclass at achieving diversification with minimal downside risk. Commodities had the least risk (smallest Max Drawdown Rebalanced and largest Sortino ratios) of Portfolios A through H.

****************************

TAKEAWAYS:
1.) Although Rick and Larry seem at the surface to be disagreeing, I think they are probably BOTH right. This really is a story of simple risk and return: commodities provided the lowest risk and lowest return. Other alternative asset subclasses required higher risk to achieve higher return.

The tradeoff I can see is:
For every 1% extra CAGR nominal required beyond Commodities, it takes at least 5.5% extra downside risk to achieve it.

2.) I believe the first debate question should not be whether Commodities are in your portfolio. I believe it should be whether individual investors diversify across Alternative Asset Subclasses. Please debate this if possible.

3.) I believe the second debate question should be how do investors tactically incrementally add alternative asset subclasses to help optimize a well diversified portfolio? It seems that low risk/low return investors would FIRST add Commodities and then add higher risk/higher return alternative asset subclasses incrementally. High risk/high return investors would perform the opposite process of FIRST adding these other subclases and adding Commodities last.

4.) We need better current tools.

If a forum member can figure out how to update the spreadsheet to enable input of the Target / Required rate of return for the Sortino ratio, that would be very helpful. It currently uses Tbills, which ain't gonna cut it for a required rate.

5.) We need better future tools.

1% additional return for 5.5% additional max drawdown seems pretty steep, but I have no data to compare it to. There must be some type of rate/return portfolio creation curve that this would fit into. I would imagine adding asset classes and subclasses probably is very productive up until a certain point. The question is where does this point sit relative to our need to take that risk (time to retirement, asset level, etc.). Although Sortino ratio is a good variable, it is essentially a FIRST derivative variable. There must be a SECOND derivative variable that you can use to understand the risk/return ratio for incremental asset class & subclass additions to a portfolio. Sortino essentially measures the rate (relative return/risk); what we need is a measure of curvature (rate change of relative return/risk).

****************************

So Larry's approach may start moving my position. However, the high cost of Commodities provide an additional return headwind that must be overcome. But let's get there when we figure out how to deal with that risk/return curve first. :wink:

CURRENT SCORE:
Commodity Alternatie Asset Subclass Allocation: NO-->Maybe (Rick 0.75, Larry 0.25)
Bond Asset Subclass Diversification: YES (Rick 1, Larry 0)
Portfolio Mirroring: YES (Rick 1, Larry 0)

Larry,
I am actually a MASSIVE fan of yours and read several of your books. You sold me on Slice & Dice. You won the asset subclass diversification war for me; this is only a post-war skirmish. :P

Thanks!!!
-Eric
Last edited by Eric White on Fri Feb 01, 2008 12:11 am, edited 11 times in total.
User avatar
tetractys
Posts: 4831
Joined: Sat Mar 17, 2007 3:30 pm
Location: Along the Salish Sea

Post by tetractys »

Third party? What would Swensen, Bogle, or Malkiel say?

Or Eric! (See above)
RESISTANCE IS FRUITFUL
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Post by larryswedroe »

Eric
Before you decide on the data try reading Ibbotson paper which does portfolio analysis. And note that Ibbotson is one of the most respected people in finance.

And also note that I have shown how adding CCF to portfolios actually enhances returns and lowers risk, leaving one with significantly higher Sharpe Ratio and of course due to lower SD better withdrawal odds of success

At my firm we have run the data using MC simulations and adding CCF improves odds of success, confirming all of the above

As an example I have shown how even during period when CCF dramatically underperformed equities they significantly improved portfolio efficiency
peter71
Posts: 3768
Joined: Tue Jul 24, 2007 8:28 pm

Re: Initial review in anticipation of the big match

Post by peter71 »

Eric White wrote:Alright folks. Just like in all the other threads, I can only read so many opinions without data before I get a little bug-eyed. :shock: Last time I looked I thought this was the VANGUARD DIEHARDS forum. WOOT! 8)

Here is the current Simba/TrevH asset allocation backtest file with candidate alternative asset subclass portfolio results:
http://groups.google.com/group/vanguard ... %2C+v4.xls

ASSUMPTIONS:
- Slice & Dice
- Value Tilt
- Alternative Asset Subclass = 10% of overall portfolio (excluding REITs)

ALTERNATIVE ASSET SUBCLASS PORTFOLIO CANDIDATES:
A = 10% Commodities (PCRIX)
B = 10% Energy Fund (VGENX)
C = 10% Health Care Fund (VGHCX)
D = 10% Precious Metals Fund (VGPMX)
E = 10% Emerging Mkts (VEIEX) ***incremental to previous input***
F = 2.5% Commodities / 2.5% Energy / 2.5% Health Care / 2.5% Precious Metals
G = 3.3% Energy / 3.3% Health Care / 3.3% Precious Metals
H = 2% Commodities / 2% Energy / 2% Health Care / 2% Precious Metals / 2% incremental to Emerging

RESULTS (CAGR Nominal / CAGR Real / Max Drawdown Rebalanced / Sortino Nominal / Sortino Nominal):
A - 14.4 / 11.0 / -11.2% / 2.3 / 1.2
B - 14.4 / 11.0 / -15.0% / 1.7 / 1.0
C - 15.0 / 11.6 / -13.5% / 2.0 / 1.2
D - 14.5 / 11.1 / -15.7% / 1.9 / 1.1
E - 15.3 / 11.8 / -16.0% / 2.1 / 1.2
F - 14.6 / 11.2 / -13.5% / 1.9 / 1.1
G - 14.6 / 11.2 / -14.3% / 1.9 / 1.1
H - 14.7 / 11.3 / -13.8% / 2.0 / 1.1

The only two portfolios to have BOTH enhanced return and low risk relative to median results are:
C
H

I personally would be a little hesitant to place all my eggs in one basket (C=Health Care), thus I would probably prefer H (even spreading of allocation across alternative asset subclasses).


DIRECT OBSERVATIONS:
1.) Rick Ferri seems correct in his claim that Commodities, relative to other alternative asset subclasses, hurt your nominal and real returns. Commodities provided the worst CAGR Nominal and Real rates of Portfolios A through H.
2.) Larry Swedroe seems incorrect in his claim that the low correlation of Commodities as an alternative asset subclass within a portfolio overcomes this impact to nominal and real returns.
3.) Larry Swedroe seems correct in his claim that Commodities are the best asset subclass at achieving diversification with minimal downside risk. Commodities had the least risk (smallest Max Drawdown Rebalanced and largest Sortino ratios) of Portfolios A through H.

****************************

TAKEAWAYS:
1.) Although Rick and Larry seem at the surface to be disagreeing, I think they are probably BOTH right. This really is a story of simple risk and return: commodities provided the lowest risk and lowest return. Other alternative asset subclasses required higher risk to achieve higher return.

The tradeoff I can see is:
For every 1% extra CAGR nominal required beyond Commodities, it takes at least 5.5% extra downside risk to achieve it.

2.) I believe the first debate question should not be whether Commodities are in your portfolio. I believe it should be whether individual investors diversify across Alternative Asset Subclasses. Please debate this if possible.

3.) I believe the second debate question should be how do investors tactically incrementally add alternative asset subclasses to help optimize a well diversified portfolio? It seems that low risk/low return investors would FIRST add Commodities and then add higher risk/higher return alternative asset subclasses incrementally. High risk/high return investors would perform the opposite process of FIRST adding these other subclases and adding Commodities last.

4.) We need better current tools.

If a forum member can figure out how to update the spreadsheet to enable input of the Target / Required rate of return for the Sortino ratio, that would be very helpful. It currently uses Tbills, which ain't gonna cut it for a required rate.

5.) We need better future tools.

1% additional return for 5.5% additional max drawdown seems pretty steep, but I have no data to compare it to. There must be some type of rate/return portfolio creation curve that this would fit into. I would imagine adding asset classes and subclasses probably is very productive up until a certain point. The question is where does this point sit relative to our need to take that risk (time to retirement, asset level, etc.). Although Sortino ratio is a good variable, it is essentially a FIRST derivative variable. There must be a SECOND derivative variable that you can use to understand the risk/return ratio for incremental asset class & subclass additions to a portfolio. Sortino essentially measures the rate (relative return/risk); what we need is a measure of curvature (rate change of relative return/risk).

****************************

So Larry's approach may start moving my position. However, the high cost of Commodities provide an additional return headwind that must be overcome. But let's get there when we figure out how to deal with that risk/return curve first. :wink:

CURRENT SCORE:
Commodity Alternatie Asset Subclass Allocation: NO-->Maybe (Rick 0.75, Larry 0.25)
Bond Asset Subclass Diversification: YES (Rick 1, Larry 0)
Portfolio Mirroring: YES (Rick 1, Larry 0)

Larry,
I am actually a MASSIVE fan of yours and read several of your books. You sold me on Slice & Dice. You won the asset subclass diversification war for me; this is only a post-war skirmish. :P

Thanks!!!
-Eric
good stuff, but i have several questions ranging from the naive to the highly technical.

1) what's the time horizon (1972-2007?) and at what interval are the risk statistics calculated -- i.e., annually over that period?

2) can you explain where if anywhere bonds are and where if ever you're testing "no alternatives other than bonds" ?

3) as hardly anyone on here knows how sortino ratios and max drawdowns are calculated can we just start with the SD's?

4) if you really like the sortino ratios, what's the MAR/target rate on which these were calculated and why's it set there. my understanding is that sortino says he likes 9%, yes . . . ? and while that's reasonable now that it's been renamed a "target" rate it just doesn't seem that falling below that captures anything i'd plausibly refer to as "downside risk."

(Edit: I see you did actually address this -- personally I like T-bills far better than anything higher like 9%, though I know some people use 0 as well so t-bills seems like a logical compromise . . . even so, because the ratios aren't as intuitive as SD I'd still like to have SD, ad I guess you'd at a minimum want to compare them to Sharpe ratios insofar as the ideal MAR is debatable).

all best,
pete
Last edited by peter71 on Fri Feb 01, 2008 12:38 am, edited 1 time in total.
Eric White
Posts: 54
Joined: Fri May 18, 2007 10:09 am

Reference materials

Post by Eric White »

Larry,

I haven't had time to research the subject in depth yet, so I appreciate your quick response. I will reserve my judgement on this preliminary data until I finish reading the threads, and read your references.

I GOTTA TRAIN UP FOR THE GREAT DEBATE!

Don't worry; I'll still be putting your book on Amazon pre-order regardless of how dry the papers may be. :lol:

I'm trying to dig up the reference from the other several threads on the subject.

Would you suggest I proceed in the following order?

1.) Ibbotson Associates, “Strategic Asset Allocation and Commodities,” March 2006
2.) Gary Gorton and K. Geert Rouwenhorst, “Facts and Fantasies About Commodity Futures,” Financial Analysts Journal (March/April 2006).

3.) Erb and Harvey paper which focuses on the "diversification return"
http://papers.ssrn.com/sol3/pa...._id=650923

Any other reerences to hit? I have an infant that's up frequently at night, so I have lots of time to read up :shock:

Thanks again, Larry. I'm looking forward to learning here and always appreciate your participation on the forum!!!

Cheers,
Eric

Thanks
peter71
Posts: 3768
Joined: Tue Jul 24, 2007 8:28 pm

Re: Initial review in anticipation of the big match

Post by peter71 »

sorry -- meant to edit the above post but i added a new one!
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

reminded by the question about bonds in the portfolio

Post by larryswedroe »

Just as reminder another factor Ferri misses in his shortsighted way of looking at things in isolation is that CCF are virtually the perfect hedge of nominal return bonds--and most high net worth people in particular hold munis

Now munis have steeper yield curve than taxables typically and thus you get rewarded more for taking that risk (and gain deflation and flight to quality hedges with longer bonds). Now by going longer you also take on more inflation risk of course. But CCF is great hedge of that risk. Not a single down long term bond year that was not accompanied by CCF being up, with average of +30% (since 1970). So when you add CCF that allows you to add maturity risk to bonds and that gets you incremental yield. That should of course be considered in the overall portfolio return if you add CCF which you would otherwise not have done without the hedge they provide.
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Post by larryswedroe »

eric
Those three are probably the best papers. Interesting idea from Erb and Harvey (from which Ferri selectively chose the point he liked and ignored the rest of the stuff). They suggest considering going long commodities in backwardation and short when in contango. That would have improved performance.
But personally I don't like that idea because you can lift your portfolio insurance at the wrong time (contango doesnt necessarily mean poor returns, just like Backwardation doesnt mean good returns).
You can also do search under Geer for papers on commodities, think he has some simple and easy to read papers.
And might try this:
Christopher A. Moth and John Kirk, “Real Return Fund: The Case for a Real Asset Class,” Institute for Fiduciary Education, July 1, 2000. And
Paul D. Kaplan and Scott L. Lummer, “GSCI Collateralized Futures As a Hedging and Diversification Tool for Institutional Portfolios: An Update,” Journal of Investing (Winter 1998).

Now see how many papers you find on the other side of this story
Last edited by larryswedroe on Fri Feb 01, 2008 12:51 am, edited 1 time in total.
Eric White
Posts: 54
Joined: Fri May 18, 2007 10:09 am

Blocking & tackling...

Post by Eric White »

Pete,

I'm the first to say how much of an amateur I am at understanding these dynamics. But I do know from my supply chain design analysis days that there is a significant difference between theoretical analysis and fact-based analysis. Theory is great; reality is better. :wink:

1.) Backtest time horizon is 23 years (1985-2007). I was stuck using this smaller horizon instead of 36 years (1972-2007) due to the lack of alternative asset subclass returns available in the longer horizon. Simbam, TrevH, and the team have been very honest and transparent about their inputs and assumptions behind this model. They would rather exclude data than corrupt it, which I appreciate. 23 years ain't great, but I'll take it over a theoretical formula anyday!

2) I excluded bonds from the candidate portfolios because I imagine the commodity risk profile (both overall volatility and especially downside risk) is more similar to equity behavior than bonds. Defining the overall bond-to-equity asset allocation is a FIRST requirement to block & tackle your risk. Then subasset classes should be suboptimized inside bonds, equities, and alternative assets separately.

I wish we had other more meaningful alternatve asset subclasses beyond those in the current backtest model. I would love to see how really noncorrelated subclasses incrementally affect the base portfolio (e.g. angels, Venture Capital, private equity, leveraged, short, short leveraged, etc.). Beside current options (real estate, Commodities) and these less likely alternatives, I'm realistically stuck with minimally correlated equity subasset classes.

3) Max drawdowns are intuititively meaningful. However, I'm a newbie at Sortino ratios. I actually went back and added max drawdowns because Sortino ratios are too academic to intuitively feel. SD's however are even worse than Sortino's, since they focus on both upside "risk" and downside risk, whereas Sortino's are solely focused on downside risk.

4) When I dug into the equations in the current model, it looked like the MAR/target rate is set at TBILLS. As I said before, that's not gonna give us an exact answer. However, they should all still scale linearly as you adjust the target rate up from TBILLS to an appropriate individual rate (e.g. 0% during accumulation phase, 4% during distribution phase for SWR).

Keep 'em comin' Pete. I want the Bogleheads to have a solid understanding of both the historical data view and theoretical view on this one. I get tired of us banging between these extremes so often on the forum.

Cheers,
Eric
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

ibbotson on commodities

Post by larryswedroe »

Eric
Saving you bit of time and hopefully helping others

Here is summary of Ibbotson paper--again one of the most respected people in finance.

Ibbotson Associates studied the evidence covering the period 1970–2004 and found the following: “Under what can only be called a very conservative commodity expected return estimate relative to historical returns the inclusion of commodities improved the efficiency (increased returns relative to risk) of a portfolio.” The paper also stated that: “This suggests that allocations to commodities do not depend on continued high returns.” Ibbotson argued that it was the low correlation to traditional asset classes that drives the allocation to commodities futures.
Using three different methodologies the authors of the Ibbotson paper found that inclusion of commodities (with allocations running as high as 31 percent) improved forward-looking returns by from 35 to as much as 77 basis points for each level of risk (standard deviation of the portfolio). Using historical returns they found the inclusion of commodities improved returns by an average of 133 basis points for each level of risk.
The authors also concluded: “Given the inherent return of commodities, there seems to be little risk that commodities will dramatically underperform other assets on a risk-adjusted basis over any reasonably long period. If anything, the risk is that commodities will continue to produce equity-like returns, in which case the forward-looking strategic allocations to commodities are too low.”

Note the key phrases, like doesnt depend on high returns for CCF, and little risk they will dramatically underperform on risk adjusted basis, etc.

And note the data would only look even better I believe if you extend it through 2007 now, and into 2008.
peter71
Posts: 3768
Joined: Tue Jul 24, 2007 8:28 pm

Re: Initial review in anticipation of the big match

Post by peter71 »

Although Sortino ratio is a good variable, it is essentially a FIRST derivative variable. There must be a SECOND derivative variable that you can use to understand the risk/return ratio for incremental asset class & subclass additions to a portfolio. Sortino essentially measures the rate (relative return/risk); what we need is a measure of curvature (rate change of relative return/risk).
i don't think the derivatives analogy holds, as you're not really interested in the rate at which the composition of your portfolio is changing. rather, it's more like a model fit problem and thus you might need something like "R-squared if item deleted," but unless I'm missing something I can't see why the difference of the "old" and "new" portfolio Sortino ratios (and/or the ratio of that difference to the old Sortino ratio wouldn't suffice).

all best,
pete


p.s. i should however add that i may have a bias against calculus in that i can barely remember it!
Last edited by peter71 on Fri Feb 01, 2008 1:27 am, edited 2 times in total.
peter71
Posts: 3768
Joined: Tue Jul 24, 2007 8:28 pm

Re: Blocking & tackling...

Post by peter71 »

Eric White wrote:Pete,

I'm the first to say how much of an amateur I am at understanding these dynamics. But I do know from my supply chain design analysis days that there is a significant difference between theoretical analysis and fact-based analysis. Theory is great; reality is better. :wink:

1.) Backtest time horizon is 23 years (1985-2007). I was stuck using this smaller horizon instead of 36 years (1972-2007) due to the lack of alternative asset subclass returns available in the longer horizon. Simbam, TrevH, and the team have been very honest and transparent about their inputs and assumptions behind this model. They would rather exclude data than corrupt it, which I appreciate. 23 years ain't great, but I'll take it over a theoretical formula anyday!

2) I excluded bonds from the candidate portfolios because I imagine the commodity risk profile (both overall volatility and especially downside risk) is more similar to equity behavior than bonds. Defining the overall bond-to-equity asset allocation is a FIRST requirement to block & tackle your risk. Then subasset classes should be suboptimized inside bonds, equities, and alternative assets separately.

I wish we had other more meaningful alternatve asset subclasses beyond those in the current backtest model. I would love to see how really noncorrelated subclasses incrementally affect the base portfolio (e.g. angels, Venture Capital, private equity, leveraged, short, short leveraged, etc.). Beside current options (real estate, Commodities) and these less likely alternatives, I'm realistically stuck with minimally correlated equity subasset classes.

3) Max drawdowns are intuititively meaningful. However, I'm a newbie at Sortino ratios. I actually went back and added max drawdowns because Sortino ratios are too academic to intuitively feel. SD's however are even worse than Sortino's, since they focus on both upside "risk" and downside risk, whereas Sortino's are solely focused on downside risk.

4) When I dug into the equations in the current model, it looked like the MAR/target rate is set at TBILLS. As I said before, that's not gonna give us an exact answer. However, they should all still scale linearly as you adjust the target rate up from TBILLS to an appropriate individual rate (e.g. 0% during accumulation phase, 4% during distribution phase for SWR).

Keep 'em comin' Pete. I want the Bogleheads to have a solid understanding of both the historical data view and theoretical view on this one. I get tired of us banging between these extremes so often on the forum.

Cheers,
Eric
Hey Eric,

Ah, our posts crossed, but as you see I did "keep 'em coming" on ever more esoteric matters as I awaited your reply :D

Re the 4 things above:

1) Sure, 1982-2007 is fine with me, though it might have implications for real vs. nominal that it's not the longer period. I'm still a bit unclear on the period of the risk stats though and one thing I like about SD is that it's generally annual SD, whereas with Sharpe or Sortino at least I don't know that that's the convention and for some reason I'm thinking maybe it's 3 years). I'm a novice with them too though so hopefully someone else will know the convention.

2) Sure thing, and I guess that as you said Larry's already sold you on alternatives in general . . . but insofar as Rick's not sold on alternatives and likes bonds etc. I just wanted to be clear that they were out.

3) Kind of dovetails on point 2) above. I personally like SD plus skewness (plus kurtosis if you like) and/or just representing things graphically. Put another way, given the problems with setting MAR and such I just don't like the idea of Sortino et al. to compress all the info available in a histogram (or even a mean, SD and a skewness coefficient) into a single non-intuitive measure. As for max drawdowns, there was a PMPT thread on here and I'll admit quickly agreeing with whoever it was who said there's something especially paradoxical about backtesting for worst case scenarios and not really thinking about it further, but I suppose I could be sold . . . how exactly are they typically calculated?

4) Agreed and this is where our posts crossed.

Thanks for your reply and all best,
Pete
peter71
Posts: 3768
Joined: Tue Jul 24, 2007 8:28 pm

Post by peter71 »

hmm, another duplicate post rather than edit . . . sorry again and either it's my computer or it's bedtime
User avatar
craigr
Posts: 2696
Joined: Tue Mar 13, 2007 6:54 pm

Re: Blocking & tackling...

Post by craigr »

Eric White wrote:1.) Backtest time horizon is 23 years (1985-2007). I was stuck using this smaller horizon instead of 36 years (1972-2007) due to the lack of alternative asset subclass returns available in the longer horizon. Simbam, TrevH, and the team have been very honest and transparent about their inputs and assumptions behind this model. They would rather exclude data than corrupt it, which I appreciate. 23 years ain't great, but I'll take it over a theoretical formula anyday!
Unfortunately this time period leaves out the 1970's high inflation which saw commodity and gold prices soar when stocks and bonds were kaput. You might want to consider running your numbers again with the 1970's included even if that means a stripped down portfolio comparison. The portfolios should then include commodities or gold as the alternative asset and see how the numbers change.

I found that alternative asset classes such as gold or commodities are going to be real dogs when stocks and bonds are doing well. But when stocks and bonds are doing poorly or inflation is a threat these alternative assets can really save your bacon as they did during the high inflation 1970's.

EDIT:

For those interested I computed the annual gold returns going back to 1972 (which is really as far as you can go anyway because gold was a fixed price before then). I got the data from here:

http://www.finfacts.com/Private/curency ... tprice.htm

Here are the numbers:

Code: Select all

Date Price   Returns
1971	43.8	0.0
1972	65.2	48.9
1973	114.5	75.6
1974	195.2	70.5
1975	150.8	-22.7
1976	145.1	-3.8
1977	179.2	23.5
1978	244.9	36.7
1979	578.7	136.3
1980	641.2	10.8
1981	430.8	-32.8
1982	484.5	12.5
1983	415	-14.3
1984	331.3	-20.2
1985	354.2	6.9
1986	435.2	22.9
1987	522.9	20.2
1988	441	-15.7
1989	433.4	-1.7
1990	423.8	-2.2
1991	379.9	-10.4
1992	356.3	-6.2
1993	419.2	17.7
1994	409.8	-2.2
1995	385.6	-5.9
1996	367.8	-4.6
1997	288.8	-21.5
1998	288	-0.3
1999	287.5	-0.2
2000	272.15	-5.3
2001	278.7	2.4
2002	346.7	24.4
2003	414.8	19.6
2004	438.1	5.6
2005	517.2	18.1
2006	636.3	23.0
2007	833.2	30.9
2008	904.8	8.6
You can go into the Simba spreadsheet and the Data_72_76 tab and put them over an asset class column of your choice. I overwrote the S-TIPS column because simulated TIPS data back to 1972 wasn't very useful to me. You can then update the column names through the spreadsheet to change S-TIPS to Gold. Gold and general commodities have very similar volatility, returns and market correlations.
Last edited by craigr on Fri Feb 01, 2008 2:01 am, edited 3 times in total.
Eric White
Posts: 54
Joined: Fri May 18, 2007 10:09 am

Followup reading

Post by Eric White »

Pete & Larry,

It looks like I have a LOT of reading ahead to catch up to both of you.

I'll try to repost this weekend when I have completed some of the background PMPT & commodities reading so I can provide meaningful responses.

Thanks for your correspondence this evening. I really appreciated it!!! :beer

Sincerely,
Eric
User avatar
Cb
Posts: 275
Joined: Wed Feb 21, 2007 12:39 am

Re: Blocking & tackling...

Post by Cb »

craigr wrote: Unfortunately this time period leaves out the 1970's high inflation which saw commodity and gold prices soar when stocks and bonds were kaput. You might want to consider running your numbers again with the 1970's included even if that means a stripped down portfolio comparison. The portfolios should then include commodities or gold as the alternative asset and see how the numbers change.

I found that alternative asset classes such as gold or commodities are going to be real dogs when stocks and bonds are doing well. But when stocks and bonds are doing poorly or inflation is a threat these alternative assets can really save your bacon as they did during the high inflation 1970's.
The further you go back the dicier the historical commodities data becomes. For the synthetic CCF dataset in the Simba spreadsheet I tried to approximate the PCRIX fund as well as possible:

2003-2007: PCRIX actuals
2001-2002: DJ_AIG plus VIPSX
1997-2002: DJ-AIG plus Yahoo IPS category returns
1991-1996: DJ-AIG plus 5-Year treasuries
1972-1990: Chase Physical Commodity Index (already collateralized)

PCRIX's substantial 0.74% ER is accounted for throughout. Look here for details:

http://gnobility.com/ER/CCF_1972-2007_san.xls

Known shortcomings:

PCRIX actuals era: Solid data, though only 5 full years

^DJC plus VIPSX era: Pretty solid, as it tracks PCRIX monthly actuals extremely closely since PCRIX inception. Remarkably, you don't even see the 0.74% ER when you regress monthly retiurns for PCRIX vs ^DJC + VIPSX (so far)...I subtract 0.74% anyway in case PIMCO loses their FI management touch

^DJC plus Yahoo IPS era: Less solid - as TIPS behaviour was unusual in the early years, but I used them anyway because the returns were slightly lower than 5Yr T's, to be conservative

Pre-1998 DJ-AIG era: Per Dow Jones: "All data prior to launch of the DJ-AIGCI on July 14, 1998 is an historical estimation using available data." Rick Ferri is suspicious of this data, thinking it is influenced by marketing. I would note, however, that D-J's 1991-1998 'estimated' data prior to going live with the DJ-AIG index averaged only 5.08% annually...not exactly a marketer's dream

^DJC plus 5Yr T-bills era: T-bills aren't a great proxy for TIPS

Pre-DJ-AIG Chase data: Raddr suggests this is probably the best DJ-AIG proxy here: http://raddr-pages.com/research/CommodityFutures.htm

I'd welcome any suggestions for improvements.

Cb 8)
ladyblunt
Posts: 38
Joined: Sat Jan 26, 2008 5:05 pm

Post by ladyblunt »

I don't know if this has been discussed before, but if one is looking strictly for negative correlation, then wouldn't adding ultrashort, or short ETFs provide another hedge to the portfolio? Of course there is no expected return by itself from a short ETF, similar to commodities, but it acts as a hedge.
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Post by larryswedroe »

One further point on Rick's comments about what I say
"Larry's mistake is that he considers low correlation as the best reason to invest in an asset class"

Not only have I never said that but I guess he quickly forgot that one of the things I have said about junk bonds is that low correlation itself is not a good reason to invest in an asset class. You have to understand the nature of asset class (in the case of junk it is hybrid, explaining the low correlation with both stocks and bonds) and also when the correlations tend to turn high and low, and one need also consider volatility and of course returns (but only in the context of how the addition of an asset impacts the risk and return of the portfolio)
peter71
Posts: 3768
Joined: Tue Jul 24, 2007 8:28 pm

Post by peter71 »

ladyblunt wrote:I don't know if this has been discussed before, but if one is looking strictly for negative correlation, then wouldn't adding ultrashort, or short ETFs provide another hedge to the portfolio? Of course there is no expected return by itself from a short ETF, similar to commodities, but it acts as a hedge.
Hi Ladyblunt,

This would add negative correlation but, as Wikipedia puts it, correlation measures the "direction" and "noise" of a relationship but not its "slope" Thus, you don't actually need the (short) "mirror image" of a high-performing (long) nvestment to get negative correlation. For example, both rows B and C below have a perfect negative correlation (pearson r = -1) with row A:

A 0,10,10,10,10
B 0,-10,-10,-10,-10
C 0,-2, -2, -2, -2

All best,
Pete
User avatar
Boris
Posts: 552
Joined: Tue Feb 27, 2007 11:10 am
Location: CT

Post by Boris »

Larry,

I don't think it's fair to start this argument on the forum (again!:)). I'd ignore it until after the debate, but that's just me.
larryswedroe wrote:And also note that I have shown how adding CCF to portfolios actually enhances returns and lowers risk, leaving one with significantly higher Sharpe Ratio and of course due to lower SD better withdrawal odds of success

At my firm we have run the data using MC simulations and adding CCF improves odds of success, confirming all of the above
My concern with this is that "CCF investing" is based entirely on history. Just like fundamental/enhanced indexing. It sure seems to have a nice spot in a portfolio based on data mining. Truth be told it's a very new investment "asset" and we don't yet know whether it gives added value (certainly not as much as we know about the stock and bond market).

I read the Groton & Rouwenhorst paper and there were several problems that stood out to me:
1) Index construction is based on data minig
2) A derivative isn't an "asset"
3) Speculative activity which depends on guessing future spot price (related to 2)
4) Survivorship bias
Page 6: "CRB database mostly contains data for futures cotracts that survived until today, or were in existance for an extended periods during the 1959-2004 period. Many contracts that were introduced during this period, but failed to survive, are no included in the database. It is not clear how survivorship bias affects the computed returns to a futures investment."
5) Selection bias
Page 6: "We based our selection on the liquidity of the contract, and it is therefore subject to a selection bias that may or may not be correlated with the computed returns."
6) Strategy bias (although page 10 shows that regular rebalancing can add to returns it's not necessary).
Page 9: "our equal-weighted index has an embedded trading strategy, which at the end of each month effectively buys a portion of those commodities that went down in price and sells a portion of those commodities that went up in price."

I'm not saying CCFs are bad, especially when considered as part of a portfolio, I'm just saying that we only have about a year or two of history, otherwise it's purely academic data-mined papers. I'm sure given full access to the CRSP database many interesting patterns can be found.

I'm still on the fence whether I'm with Bogle and total markets or whether I'm with Fama/French and small/value premium, so what do I know :). I have my own ideas that I'd rather not go into in this thread.

Looking forward to the debate.

Boris
Short term moves in the market are like "a tale Told by an idiot, full of sound and fury, Signifying nothing." | - John C. Bogle quoting Shakespeare
User avatar
Boris
Posts: 552
Joined: Tue Feb 27, 2007 11:10 am
Location: CT

Post by Boris »

peter71 wrote:
ladyblunt wrote:I don't know if this has been discussed before, but if one is looking strictly for negative correlation, then wouldn't adding ultrashort, or short ETFs provide another hedge to the portfolio? Of course there is no expected return by itself from a short ETF, similar to commodities, but it acts as a hedge.
Hi Ladyblunt,

This would add negative correlation but, as Wikipedia puts it, correlation measures the "direction" and "noise" of a relationship but not its "slope" Thus, you don't actually need the (short) "mirror image" of a high-performing (long) nvestment to get negative correlation. For example, both rows B and C below have a perfect negative correlation (pearson r = -1) with row A:

A 0,10,10,10,10
B 0,-10,-10,-10,-10
C 0,-2, -2, -2, -2

All best,
Pete
Actually, it's a good point.. what if you invest the way ladyblunt said and see if you can find a 'rebalancing' premium? :) Stock market + protection using derivatives ~= SD/return of treasuries, right? I'm no Sharpe, but I bet it's arithmetically accurate, in theory. What happens if you constantly rebalance? Is there a rebalancing premium on top of the "risk-free" returns? :) Now there's a research paper for someone!

Boris
Short term moves in the market are like "a tale Told by an idiot, full of sound and fury, Signifying nothing." | - John C. Bogle quoting Shakespeare
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Post by larryswedroe »

Boris
let's address those issues
read the Groton & Rouwenhorst paper and there were several problems that stood out to me:

1) Index construction is based on data mining
Data mining is to mine the data and then create a hypothesis. For example finding a specific period that gives you a certain result.
This was not the case here. They used to longest data period available, which is very long for investing. Longer for example than the period used for Fama and French's famous paper

2) A derivative isn't an "asset" --that is not really relevant. There is an underlying asset. Derivates are not good or bad, just a way of accessing risks and expected returns.

3) Speculative activity which depends on guessing future spot price (related to 2) --that is not true. No guess is made. The only assumption is IMO that commodities will move in line with inflation. And that is historically the case. And also that they generally react positively to event risks. Example, more likely to crisis from drought than from outbreak of good weather. Or a war and how it impacts prices. Remember commodities positively correlated to inflation while reverse true for stocks and bonds.
4) Survivorship bias
Page 6: "CRB database mostly contains data for futures cotracts that survived until today, or were in existance for an extended periods during the 1959-2004 period. Many contracts that were introduced during this period, but failed to survive, are no included in the database. It is not clear how survivorship bias affects the computed returns to a futures investment."
I dont see this as an issue really, it is not like with stocks where the stocks go to zero and there is bias. Just because a futures contract may have disappeared doesnt tell you the same thing. The price of the commodity obviously did not go to zero.
5) Selection bias
Page 6: "We based our selection on the liquidity of the contract, and it is therefore subject to a selection bias that may or may not be correlated with the computed returns."

Yes this is possible though I don't think it matters that much based on studies on various weightings showing similar long term results.

6) Strategy bias (although page 10 shows that regular rebalancing can add to returns it's not necessary).
Page 9: "our equal-weighted index has an embedded trading strategy, which at the end of each month effectively buys a portion of those commodities that went down in price and sells a portion of those commodities that went up in price."

I dont really consider a regular rebalancing to be a trading strategy--it is in fact not rebalancing that would be one. You establish an asset allocation and you stick to it, not letting market determine your AA. More importantly the question should be is does this make for intelligent design of an "index" or investment. The answer is yes because it takes advantage of the high volatility of the individual commodities and their very low correlation, leading to a large diversification return. This is the point Erb and Harvey made and state that this is the most reliable part of the return. And it is large.

I hope that helps
peter71
Posts: 3768
Joined: Tue Jul 24, 2007 8:28 pm

Post by peter71 »

Boris wrote:
peter71 wrote:
ladyblunt wrote:I don't know if this has been discussed before, but if one is looking strictly for negative correlation, then wouldn't adding ultrashort, or short ETFs provide another hedge to the portfolio? Of course there is no expected return by itself from a short ETF, similar to commodities, but it acts as a hedge.
Hi Ladyblunt,

This would add negative correlation but, as Wikipedia puts it, correlation measures the "direction" and "noise" of a relationship but not its "slope" Thus, you don't actually need the (short) "mirror image" of a high-performing (long) nvestment to get negative correlation. For example, both rows B and C below have a perfect negative correlation (pearson r = -1) with row A:

A 0,10,10,10,10
B 0,-10,-10,-10,-10
C 0,-2, -2, -2, -2

All best,
Pete
Actually, it's a good point.. what if you invest the way ladyblunt said and see if you can find a 'rebalancing' premium? :) Stock market + protection using derivatives ~= SD/return of treasuries, right? I'm no Sharpe, but I bet it's arithmetically accurate, in theory. What happens if you constantly rebalance? Is there a rebalancing premium on top of the "risk-free" returns? :) Now there's a research paper for someone!

Boris
Hi Boris,

I guess it would depend on the percentages of your short hedge, but unless i'm misunderstanding how those short ETF's work, buying a 50% S&P ETF and a 50% S&P short ETF is going to get you a return of 0 minus transactions costs, no? I know no one is advocating 50/50, but the point is that even if one is proposing 80/20 there's probably some better way for buy and holders to go about looking for negative correlations.

All best,
Pete
User avatar
Cb
Posts: 275
Joined: Wed Feb 21, 2007 12:39 am

Post by Cb »

Boris wrote:
I'm not saying CCFs are bad, especially when considered as part of a portfolio, I'm just saying that we only have about a year or two of history, otherwise it's purely academic data-mined papers. I'm sure given full access to the CRSP database many interesting patterns can be found.

Boris
Boris, we have a bit more than a year or two of CCF history. Have a look at this:

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

There are now 60 months of PCRIX returns available afrom Yahoo (I don't know why they don't include the first 6 months of the fund's existence)

Based on it's strong (0.998) correlation to ^DJC plus VIPSX I think it's reasonable to append another 30 months of DJC + VIPSX. Do you agree?

I'm going to download monthly data for TM, Int'l TM, Bonds, REITs etc to see how those 90 months of CCF returns correlate to them when I get time.

Cb
User avatar
Boris
Posts: 552
Joined: Tue Feb 27, 2007 11:10 am
Location: CT

Post by Boris »

peter71 wrote:Hi Boris,

I guess it would depend on the percentages of your short hedge, but unless i'm misunderstanding how those short ETF's work, buying a 50% S&P ETF and a 50% S&P short ETF is going to get you a return of 0 minus transactions costs, no? I know no one is advocating 50/50, but the point is that even if one is proposing 80/20 there's probably some better way for buy and holders to go about looking for negative correlations.

All best,
Pete
Peter,

My comment was somewhat meant in jest, however I didn't mean that one invest directly in that particular ETF. What I'm suggesting is that if one invested in a security and used options to protect against the downside risk (insurance) then the closer they got to a "risk free" return (not including transaction costs), the closer they'd get to the returns of treasuries (definition of "risk-free").

Now, I'm not sure if I can even state the sentence above, certainly it's not based on any of my studies of Economics, it's just something that seems to make sense. Based on that statement I'll extrapolate and suggest that: This type of investing does generate a positive return (~equal to treasuries), with unknown costs. I will then relate this to investing in commodities, which has expected return of inflation, however due to volatility and negative correlation can produce a significant rebalancing bonus. I will then state that because investing in a particular security and buying insurance for that security are negatively correlated, then why wouldn't the same thing that is true for commodities not true in the case I've stated?

I can see a few problems with my assumption, namely that "all" commodities go up with inflation, which isn't true for the case I presented, however that's why I said this maybe a good research topic for someone (unless I've just made an obvious mistake in my assumptions).

I don't know what the exact ratio should be, it should be enough to make the equities return equal to risk-free treasuries. I don't know how or when you should rebalance, that requires data-mining.

Boris
Short term moves in the market are like "a tale Told by an idiot, full of sound and fury, Signifying nothing." | - John C. Bogle quoting Shakespeare
lasttoknow
Posts: 93
Joined: Sun Nov 11, 2007 7:19 pm

Post by lasttoknow »

I have a quick question for you experts. How does a money supply growing at 15% rate affect the prices of commodities versus stocks and bonds? Also, in the event of hyperinflation (Chile, Argentina, Russia, China, Germany, France, Greece, Hungary, Serbia, Bolivia, etc), what is the relative tradeoffs between owning bonds, equities, and commodities? Thanks.
peter71
Posts: 3768
Joined: Tue Jul 24, 2007 8:28 pm

Post by peter71 »

Boris wrote:
peter71 wrote:Hi Boris,

I guess it would depend on the percentages of your short hedge, but unless i'm misunderstanding how those short ETF's work, buying a 50% S&P ETF and a 50% S&P short ETF is going to get you a return of 0 minus transactions costs, no? I know no one is advocating 50/50, but the point is that even if one is proposing 80/20 there's probably some better way for buy and holders to go about looking for negative correlations.

All best,
Pete
Peter,

My comment was somewhat meant in jest, however I didn't mean that one invest directly in that particular ETF. What I'm suggesting is that if one invested in a security and used options to protect against the downside risk (insurance) then the closer they got to a "risk free" return (not including transaction costs), the closer they'd get to the returns of treasuries (definition of "risk-free").

Now, I'm not sure if I can even state the sentence above, certainly it's not based on any of my studies of Economics, it's just something that seems to make sense. Based on that statement I'll extrapolate and suggest that: This type of investing does generate a positive return (~equal to treasuries), with unknown costs. I will then relate this to investing in commodities, which has expected return of inflation, however due to volatility and negative correlation can produce a significant rebalancing bonus. I will then state that because investing in a particular security and buying insurance for that security are negatively correlated, then why wouldn't the same thing that is true for commodities not true in the case I've stated?

I can see a few problems with my assumption, namely that "all" commodities go up with inflation, which isn't true for the case I presented, however that's why I said this maybe a good research topic for someone (unless I've just made an obvious mistake in my assumptions).

I don't know what the exact ratio should be, it should be enough to make the equities return equal to risk-free treasuries. I don't know how or when you should rebalance, that requires data-mining.

Boris
Hi Boris,

Sure, I think absolute return funds use covered calls, protected puts, etc., but, as you say, I think they're for the most part more interested in literally insuring or hedging positions than in mean-variance optimization. I'm no expert on hedge funds, though, so others will likely know more. Perhaps more important, though, the risk free return is by definition just that, so I can't see any real incentive for any fund (including absolute return funds) to go through all these machinations and costly transactions just to equal the return on T-bills.

All best,
Pete

P.S. Per lasttoknow's question, I'm an academic but not an economist so I have no idea, but I know various measures of money supply vs. various measures of price increases is something at least some economists fight about . . . As for stocks/bonds/commodities under conditions of hyperinflation, I guess you could set a floor of 15% or so for hyperinflation and look at the historical data, but I certainly haven't done so myself.
Post Reply