In my last regular blog post here, I compared two utilities ETFs (JXI for international and IDU for domestic) to a model portfolio that combined JXI with a number of individual utility stocks. The purpose of the comparison was to show that it was straightforward to identify a portfolio of utilities with higher yield than the ETFs, about the same volatility, and neutral exposure to Treasury yield.
The exposure to Treasury yields is measured using the correlation between the returns of a portfolio or individual stock or ETF and the 10-year Treasury yield. This is a very important variable because utilities tend to be somewhat interest rate sensitive. In addition, given the current environment, it is worth paying careful attention to how your portfolio may perform in the face of rising rates.
In my article, I noted that there was a substantial difference between the correlation of JXI to 10-year Treasury yield using three years of historical data vs. give years. Given this, I chose to use a five year period as the basis for comparison between my model portfolio and the utility ETFs. This discussion may have caused some confusion. One comment asked why I would think it is meaningful to use the correlation to 10-year Treasury yield as a guiding variable if it changes so much with the sample historical period. This issue deserved a good answer.
The same question may be leveled at the use of any historical statistical metric. Given that these change in time--correlations are notoriously unstable, but volatility is also quite variable--does it really make sense to use these metrics to guide portfolio construction. To be sure, care must be taken. In the table below, I show statistics for yield along with volatility over 3-, 5-, and 7-year periods as well as correlations of portfolio returns to 10-year Treasury yields over these same periods.
|Portfolio||Yield||Historical Volatility||Correl. to 10-Yr Treas. Yield|
|3 years||5 years||7 years||3 years||5 years||7 years|
|Model Utility Portfolio||5.3%||10.5%||13.5%||14.8%||2.6%||0.0%||2.5%|
|Equal wt. to JXI, XLU, IDU||3.7%||10.2%||12.6%||14.1%||-11.2%||-5.3%||-2.0%|
While the values vary in time, the relative relationships are quite consistent. In particular, the model utility portfolio from my previous post has higher yield than JXI or a mix of JXI, IDU, and XLU, with about the same level of risk and a correlation to 10-year Treasury yield that ranges from very slightly positive to zero.