Correlation Diversification Reduces Portfolio Risk

Editor's Note: Please see 36-Month ETF Correlations with Russell 3000 for corrections on some of the data found in this article. Diversifying across growth or value styles, market capitalizations, regions or countries does not necessarily provide risk protection against the components of your portfolio moving up or down at the same time or to the same degree. Diversifying by correlation does help prevent all your portfolio components from marching in unison.

Correlation is a statistical measure of how two securities move in relation to each other. Correlation is expressed by numbers ranging from -1 to +1. Perfect negative correlation means the two securities move lockstep in opposite directions. Perfect positive correlation means the two securities move lockstep in the same direction. Zero correlation means the two securities move randomly with respect to each other.

Watching your portfolio soldiers marching shoulder-to-shoulder in an up market is a lovely thing, but that sets you up for them to march together in a down market – and that’s an ugly thing.

Portfolios diversified as to correlation generally don’t go up as fast or far as non-diversified portfolios and they generally don’t go down as fast or far either.

The chart below shows the correlation between the broad US stock market and the growth and value styles for the whole market, for large-cap stock and for small-cap stocks … no hiding places there.

ETFs shown are:

• (IWV) – Russell 3000
• (IWW) – Russell 3000 Value
• (IWZ)– Russell 3000 Growth
• (IWB) – Russell 1000
• (IWD) – Russell 1000 Value
• (IWF) – Russell 1000 Growth
• (IWM) – Russell 2000
• (IWN) – Russell 2000 Value
• (IWO) – Russell 2000 Growth

The next chart shows that little in the way of correlation diversification is available by spreading assets between domestic and foreign stocks, whether in developed or emerging markets. It does show that bonds provide diversification through negative correlation.

The chart expresses the correlation of each ETF to the broad US stock market as represented by the Russell 3000 through its proxy ETF, IWV. The other ETFs are:

• (EFA) – MSCI EAFE (Europe, Australasia, Far East)
• (EEM) – MSCI Emerging Markets
• (AGG) – Lehman Aggregate Bond Index
• (IEF) – Lehman 7-10 Year Treasury Bonds

The final chart shows the correlation of key US industrial sectors versus the broad US stock market (Russell 3000).

Good luck trying to hide from correlation risk with these. The Russell 3000 is represented by IWV. The other ETFs are:

• (IYM) – Basic Materials
• (IYC) – Consumer Services
• (IYK) – Consumer Goods
• (XLE) – Oil & Gas *
• (IYF) – Financials
• (IYH) – Health Care
• (IYJ) – Industrials
• (IYW) – Technology
• (IYZ) – Telecommunications
• (IDU) – Utilities

We did not use the Dow Jones Oil & Gas sector ETF, because it did not have enough history to calculate the correlation. We substituted the Energy SPDR ETF as a reasonable stand-in.

We’ll look at other asset classes in another article to seek more opportunity other than bonds (or cash) to diversify correlation risk.

Comments

[david cohen_:]

Very interesting article. The high correllation of so many asset classes (particularly foreign and US stocks) is somewhat different than I would expect. Iwonder what the correlations with REITs would be since they are touted as good portfolio diversifiers. Thanks.

2007 Jan 04 12:00 PM

[QVM Group:]

Thanks David. I will answer that question. Stay tuned. My plan is to run through several categories of securities with this test. I have just finished on on gold mining stocks that may be published and will be submitting another on different types of bonds shortly. You have a good idea on REITS. There may be some difference based on the underlying property types, but my assumption (and I'm doing these tests to exchange assumptions for observations of fact) is that REITS have a hybrid quality somewhere between bonds and stocks in their correlation. It might be interesting to test them side-by-side with other vehicles with similar distribution and taxation characteristics, such as master limited partnerships and royalty trusts too -- how much does the major dividend feature set them all apart?

2007 Jan 05 01:19 AM

[QVM Group:]

Quick answer David --- looking at the REIT ETFs, the had a 1-year correlation around +0.85 to +0.88 for 12 months and around +0.95 for 3 years. No hiding place there.

2007 Jan 05 08:50 AM

[QVM Group:]

David, I have posted a major correction to this article which you should read given your surprise at the correlations in this article. I was surprised to, but the source was an issue. The new article explains. Please do read it. etf.seekingalpha.com/a...

2007 Jan 12 12:21 AM

[market man:]

QVM Group - take a look at the correlation difference of a traded REIT and a non traded REIT. I believe you will see a difference - the traded REIT will have a higher correlation than non traded.

Jul 30 05:58 PM