Using Non-Correlating Asset Classes

[sengsational]

This is an interesting article about why it might be worthwhile to go a bit farther than "X percent stock and Y percent bonds" in your asset allocation. There's a lot of discussion about it because correlation is so variable.

Historical correlations can rise and fall making the science of which indexes belong together in the same asset class more of an art. Complicating matters, correlations on most stock indexes often approach 1.00 during times of market turmoil.

Over the last 3 years, iShares North America Natural Resource ETF (IGE) had a correlation of .68 with the S&P 500 and Vanguard’s US REIT (VNQ) had a correlation of about .31 (compared to over .90 if you "diversify" from the S&P 500 with small cap stocks, for instance).

Do Resource Stocks Deserve Their Own Asset Class? | Marotta On Money

 

[kevink]

Don't look at the past 3 years, look at what happens during market crashes like the one we had in '08 when all of those supposedly non-correlated asset classes tanked in unison. If you held a portfolio of truly non-correlated assets such as Harry Browne's Permanent Portfolio (25% each U.S. Stock Index, 30 Year Treasuries, Treasury Money Market Fund and Physical Gold) you did well - if not, you got killed.

Physical gold beats the hell out of resource stocks in a market panic, and so does the flight to safety that only U.S. treasuries provide. Not that the PP is by any means the only way to do this - Larry Swedroe's "no fat tails" approach (75% 5 year Treasuries, 25% small cap value stocks divided equally between the U.S. and international) is nearly as good.

 

[sengsational]

Good points. I don't necessarily agree that your AA should be based only on the on the worst crash event, but instead a time period that includes the worst crash. The length of that time period should be as long as your non-volatile holdings would last you. I'll write the author and see if we can get a blog post on correlation during market panics.

I liked the "Using Non-Correlating Assets..." post, but got to wondering how those asset classes would have correlated in 2008. It seemed like nothing held value then, pushing the correlation toward 1.00, no matter what you were invested in.

It's probably not useful for the long-term investor to know correlation between asset classes just for the time period including the big slide downwards, but if an investor had, say, X years of non-volatile assets to weather the storm, then correlation over a span of X years that included a panic might tell an more nuanced story?

 

[Lsbcal]

I'm not so sure about taking one segment of the stock market and looking at correlation to the general market over a few years. Owning a broad based index fund seems smarter then cutting it up and emphasizing segments.

Why not a value tilt which ought to include the temporarily lower correlating and lower performing segment? Plus a value tilt might tend to include that underperforming segment and exclude it when it outperforms for a long period. I don't have any good data to back up that last sentance but I do tend towards value tilting.

 

[papadad111]

Don't look at the past 3 years, look at what happens during market crashes like the one we had in '08 when all of those supposedly non-correlated asset classes tanked in unison. If you held a portfolio of truly non-correlated assets such as Harry Browne's Permanent Portfolio (25% each U.S. Stock Index, 30 Year Treasuries, Treasury Money Market Fund and Physical Gold) you did well - if not, you got killed.

.

Bingo. Nailed it. +1. Even real estate is supposed to be non- correlated to equities and look what happened. Commodities too. Kaboom.

 

[mathjak107]

each time is different enough to make new bedfellows.

last time long term treasuries soared when equities tanked.

but if rising rates kick off the next round and inflation picks up bonds may be chopmeat and commodities the new bedfellow.