.
.
.

Larry, Bill Coaker did find assets which have been CONSISTENTLY low correlated and emerging markets and international stocks were amongst them.

We have discussed it before, see this:

fpanet.org/journal/articles/2007_Issues/upload/56354_1.pdf


Excerpt: “International stocks, emerging markets, real estate, natural resources, and long-short have consistent lower correlations to US equities”.

As to International Small Cap Index funds, the problem is that Vanguard International Explorer VINEX is closed to new investors. So the alternatives are DFE, GWX, etc. or from here:

biz.yahoo.com/p/tops/fa.html

Not particularly low costs.
.
.

Vig
there is no such thing as an equity asset that has consistently low correlation. They have low AVERAGE correlations.
Just look at periods of crisis and you will find they generally all coverage in terms of correlations. Which is why the best diversifiers are fixed income and assets with negative correlation, of which there are basically two, TIPS and CCF. And even they are not perfectly negative. In 25% of the negative years for stocks CCF produced significant losses.
Which is why I had significant issues with the paper you cite

.
.

Hi Larry. I get the following Risk Impacts when I enter these asset classes into Bill Coaker’s Balanced Low Correlated portfolio (on pg 68 of the PDF, but it could be other portfolios), at the www.Riskgrades.com website:

{Values of RiskGrades are in brackets [RG] }

Vanguard Total Bond Market index VBMFX [32] and T. Rowe Price international bonds RPIBX [43] ----at minus (-4%).

Vanguard TIPS VIPSX [44]--at minus (-2%)

PIMCO Commodity PCRIX [128] --at plus +2%.

“Risk Impact of an asset = measures how your portfolio's RiskGrade would change if the position was removed-that is if the position was closed out and the proceeds kept in cash.”

So Larry, they are saying that at this time intermediate nominal US or foreign bonds have lower correlations to US equities than TIPS and CCFs.

You can read their Technical Document here:

THE RISKMETRICS® GROUP, LLC
RISKGRADES™ TECHNICAL DOCUMENT
USER LICENSE AGREEMENT

www.riskgrades.com/retail/clients/dnldtechdoc/
.
.
p.s. Bill Coake's above mentioned portfolio shows good diversification benefits by lowering the portfolio's risk by 35%.
.
.

.
.
Larry, if I enter onto the RiskGrades website a recent Gene Fama, Jr. suggested “Global Exposure” allocation, I see that the VBMFX (Vanguard total bond market) has a Risk Impact of minus -4% on the portfolio while the TIPS VIPSX has more at minus (-6%).

BTW, the Vanguard International Small VINEX has a +10% Risk Impact (bad) and the Vanguard International Value VTRIX has + 15% (even worse).

The portfolio:

30% S&P 500 VFINX
8% US Small Cap index NAESX
10% US Lrge Cap Value index VIVAX
10% International Small Cap VINEX
10% International Value VTRIX
32% Lehman Gov/Credit (replaced with Vanguard VBMFX)

Fama entered a 14.07% for annualized return at 10.09% SD for the above allocation, representing the market from 1975 to 2005.

Vig
My only commentg would be that if you get answers like that the data is nonsense, shows how dangerous data in the wrong hands can be. There is simply no way the TBM is even close in risk to TIPS.

MY assumption is the model is bad, it is way too reliant on short term data. Garbage in and garbage out

Thanks Larry, here is Bill Reichenstein’s opinion on the RiskGrades (to me from 2003)

"RiskGrades"

"My analysis is primarily based on material in the RiskGrades Technical Document.

In a nutshell, there is much to be said for their risk measures, but I have strong objections to their return measures and, therefore, risk-adjusted return measure.

They estimate future returns from a relatively recent returns history. The history is 2 to 5 years for some assets and at least one business cycle for others. The expected return on Japanese stocks and tech stocks is positive despite their recent historic returns. DeBondt and Thaler among others show that individual stocks that perform best in one three-year (or five-year) period tend to perform worse in the next three (or five) years. This is the long-run return reversal effect. The Technical Document says "historic return is not necessarily a good forecast." I agree. It also says "because we lack a reasonable expected return computation, almost by default we use a longer-term historic return as the benchmark. If an investor has found a better measure of expected returns, we strongly recommend using it" (p. 41). If you selected stocks based on their risk-return measure (Eq 5.2), you would do little more than select the best performing stocks over the recent past. Differences in recent historic returns are HUGE. Differences in recent standard deviation (or their RiskGrade) are relatively small. Thus their method would have a strong bias toward recent stock winners. It would have recommended tech stocks in early 2001 and said to avoid them now. Value Line is one source of expected return that is forward looking and not backward looking. Perhaps its mid-point of 3-to-5 year return (after annualizing) could be used as a measure. But I do not recommend the ReturnGrade.

The RiskGrade is similar to standard deviation except 1) it places more weight on recent days and 2) it is standardized so a grade of 100 is the risk of an average stock index from 1995-99. The decay factor of 0.97 implies use of 151 days of historic returns. About 60% of a year or seven months. This risk measure thus varies through time. RiskGrades will be high shortly after volatile periods-Asian flu and Russian default. In my opinion as a long-term investors, the ex ante or expected long-run volatility is less volatile than RiskGrades; in essence, I place weight on return volatilities over a much longer historic period. Recognizing the limitations due to its emphasis-right or wrong-on the recent past, I still believe the RiskGrades approach has much merit. Volatility-whether prices are moving up or down-implies that prices will likely continue to be volatile. "

A historic risk chart would show the volatility over rolling 151-day periods. Something like the Asian flu would reasonably quickly produce a noticeable increase in RiskGrades. And that spike would dissipate through time. How good is the historic risk chart? In my opinion, the key issue is how well the past 151 days depicts ex ante risk for your investment horizon. After periods such as the Asian flu, I am not convinced long-run stock market risk is substantially above average. Similarly, after months of relative calm, I am not convinced long-run risk is substantially below average."
*******************************
As to the words "CONSISTENTLY low correlation" used by Bill Coaker here is an example: "Natural Resources (NR) have had a correlation of less than .20 to all 17 other assets in the study (18 asset classes in total but NO TIPS) with the highest being just .19, for both small growth and small value. NR have had the lowest average correlations-and the most consistently low correlation-to every asset in the study, including every category of stocks, bonds, and alternative. Hence, NR have provided more more diversification benefits than every other asset in the study" (1970 to 2007).
******************************
Larry, since Coaker used Goldman Sachs Commodity Index for NR, you both are saying the same thing. But Coaker continued with using the words "CONSISTENT low correlation" for Long-Short, US bonds, and Global bonds.

p.s. question: how are you doing with the RETIREMENT book?

Best.




The current lay- of- the- land at the RiskGrades website. A sample of US funds with less risky at top. Note the funds inception dates.

(DFIHX) DFA Investment Dimensions Group Inc: One-Year Fixed Income Portfolio; Institutional Class Shares RiskGrade: 2
Min: 2 Max: 9 Avg: 4 (08/02/2000-04/17/2008)
------------------------------------
(VBISX) Vanguard Bond Index Funds: Vanguard Short-Term Bond Index Fund; Investor Shares RiskGrade: 19
Min: 7 Max: 20 Avg: 11 (08/02/2000-04/17/2008)
-----------------------------------------
(VBMFX) Vanguard Bond Index Funds: Vanguard Total Bond Market Index Fund; Investor Shares RiskGrade: 31
Min: 11 Max: 34 Avg: 18 (08/02/2000-04/17/2008)
----------------------------------------
HSTRX) Hussman Inv:Strat TR RiskGrade: 33
Min: 18 Max: 40 Avg: 29 (03/03/2004-04/17/2008)
----------------------------------------
(VIPSX) Vanguard Fixed Income Securities Funds: Vanguard Inflation-Protected Securities Fund; Investor Shares RiskGrade: 44
Min: 13 Max: 49 Avg: 26 (01/02/2001-04/17/2008)
------------------------------------

The TIPS's RiskGrades according to them, only in 3Q 2003 and after mid 2007 rose to a RG of above 40.

Larry, have you changed your mind that it is not GIGO?

Vig
No it is GIGO IMO,
as I said you can simply look at the data yourself--there is no equity asset class that has persistent low correlation to other equities--just look at any period of crisis and you can see that.
Correlations drift and are just average figures--they dont reflect the fact that during crisis all correlations can rise--of course at the wrong time.
And just the one example we looked at shows how silly this stuff is IMO.

the international data you use is unreliable to say the least. Many eafe countries including france italy and spain had capital controls well into the 1980s restricting foreign investment and capital flows. The same holds for developing markets well into the 1990s.Based onyour background Larry S I am sure you know this far better than I do.

To put it simply international investing in the 1970s was far different than in 2008 this is only one of the reasons add growth of volume, growth of instruments, more liquidity and it will be of little surprise correlation is basically in a secular trend upwards and as they say nothing goes up in a down mkt other than correlations.

data on international mkts and their correlation to the US from the 1970s and most of the 1980s is GIGO in drawing conclusions about today's mkts

check the data on the rolling 5 yr correlations since the 1970s