Country vs. Sector Diversification with ETFs

Some time ago when I was in graduate school I had a class entitled “Global Portfolio Management”. Aside from doing rigorous things like calculating beta and portfolio standard deviation BY HAND, there were a few interesting and salient points that stuck with me. One of those happened to concern the relationship of price movements between two classes of stocks; 1) all of the stocks within a given country but in separate industries and 2) all of the stocks within a given industry but in separate countries.

Now, I cannot find the original academic work that my professor relied upon to demonstrate this point but the conclusion of the work went something like this (paraphrasing):

The movements of price returns for stocks (in all industries) within a single country are more correlated with one another than the movements of price returns for stocks (in all countries) within a single sector/industry.

Anyone who has done any sort of portfolio construction realizes that the correlations of the individual asset’s price returns to one another is what dictates the degree of diversification within the portfolio. So combining those two ideas, IF stocks within the same country are more correlated to one another than stocks within the same industry, we would expect the degree of diversification within an individual country to be less than within an individual industry. We would also then expect the volatility (prices would move more in line with each other) of owning individual countries to be higher than owning industries/sectors. To test this idea I plugged 20 of the iShares MSCI individual country ETFs and 10 iShares global sector ETFs into QPP and ran historical tests and forecasts.

The look back period was 3 years for all the funds that had that much data but slightly less than that for iShares S&P Global Consumer Discretionary Sector ETF (RXI), Global Utilities (JXI), Global Materials (MXI), Global Consumer Staple (KXI) and Global Industrials (EXI) which have been around less time. The results seem to confirm the above hypothesis.

A few observations. The averages at the bottom of both tables have two lines, the first is for ALL the ETFs in the table. The second is a sort of smoothed average of all the ETFs not shaded in yellow. I removed these because they had the highest standard deviations for the sets and I wanted a second comparison without this skew. Both sets seem to provide the same conclusion. The countries removed were Brazil, Mexico, and South Africa (Emerging Markets). The sector removed was Energy.

Notice that the average Beta and Standard Deviations are higher for the country ETFs than for the sector ETFs. Notice also that the forward looking return estimates (the Return column) consider this. This data does seem to confirm the notion that stocks within a country tend to be less diversified than stocks within a sector (and hence more risky).

The conclusion from this is pretty simple. When constructing a portfolio be mindful of the fact that individual country ETFs will tend to have a higher risk (and therefore expected return) profile than individual sector ETFs. Make sure your portfolio analysis considers this difference in potential volatility.

Comments

[granger]

Great info, thank you. I loved it.

[lazytrader]

Foreign ETFs are big contributors to my balanced portfolio, however, what I find more often than not is that many of these ETFs have high concentrations on specific sectors - e.g. financials or oil & gas, etc. which leads to eliminating them from a balanced portfolio consideration due to high correlation to other domestic ETFs. ILF is one of the more profitable components I've been using recently. See my full portfolio at notiming.com