In a recent post, I wrote that utilities—as an asset class—are not currently offering enough yield to justify their risk levels, although individual utilities may be worth considering. This begs the question as to which utilities look relatively more attractive than the sector as a whole and how we might determine this.
In my approach to designing income portfolios, I focus on two metrics: yield and risk. Yield is directly observable and risk is observable in hindsight but tends to have a high level of persistence. I am not going to go into depth here, but I have written a number of articles comparing historical risk levels, projected risk levels, and option implied volatility. They tend to be remarkably consistent. My approach to calculating a projected risk combines trailing risk for an asset—say the iShares U.S. Utility ETF (IDU)—with a model that adjusts this volatility based on the correlation to the broader market and the market’s volatility. For this discussion, I am just going to look at trailing risk (volatility) and yield to simplify the discussion.
At the end of November 2013, IDU has a yield of 3.2% and a trailing 3-year volatility of 10.6%. The iShares Global Utilities ETF (JXI) has a yield of 4% and a trailing 3-year volatility of 11.4%. The relative yield vs. risk seems consistent. JXI provides a slightly higher yield relative to its volatility level. There is however a big difference in interest rate sensitivity. IDU has a -20.8% correlation to the 10-year Treasury bond yield, while JXI has a +17.5% correlation to the 10-year Treasury yield over the past three years. This means that JXI has a modest tendency to rise when Treasury yields rise, while IDU displays the opposite tendency. These correlations are both quite small in magnitude, of course, and suggest that other factors tend to dominate. One oddity in the data is that the trailing five-year correlation between JXI and 10-year Treasury yield is 0%, as compared to +17.5% over the three-year period. This suggests that the correlations over a given period may be largely determined exogenous factors and thus are somewhat unstable. IDU has a -6% correlation to the 10-year Treasury yield over the five-year period.
Another major distinction between JXI and IDU is their trailing performance. IDU has returned an annualized 0.36% per year over the past five years as compared to 7.45% per year for IDU over the same period.
These yield and risk levels for the utilities ETFs are such that they do not get included in the optimal yield portfolios—an efficient frontier of yield vs. risk. There are individual asset classes with far more attractive properties. MLPs (AMLP) and preferred share funds (PFF) provide more yield with less risk and a higher positive correlation to the 10-year Treasury yield. In the current environment, a higher positive correlation to changes in Treasury yield is a very desirable trait.
The question is whether some combination of individual utilities can be found that is more attractive than the utility ETFs. In my recent post linked above, I suggested that this was the case. Here is why.
I have screened the universe of utilities, with some filters for maximum volatility and minimum yield. We are looking for solid income providers but we want to avoid companies in distress which are more likely to cut their dividends.
|Name||Ticker||Yield||Correlation to 10-year Treasury Yield||Trailing 3-year Volatility|
|American Electric Power||AEP||4.3%||-37.4%||13.5%|
|Empire Dist. Electric||EDE||4.5%||4.4%||14.2%|
|iShares U.S. Utilities ETF||IDU||3.2%||-20.8%||10.7%|
|iShares Global Utilities||JXI||4.0%||17.5%||11.5%|
I have also included the two utility ETFs (IDU and JXI) in the table. The correlations to the 10-year Treasury bond yield are over the past three years.
The variations in the risk levels is notable, but nothing new. Certainly investors in First Energy (FE) should be aware that the very high yield they are getting is compensation for a high level of risk. The yield of FE is about twice that of IDU and the risk level of FE is also about twice that of FE. This sort of thing makes the markets look fairly rational.
The question that we are really interested in is whether we can create a portfolio of these stocks–and perhaps some allocation to the ETFs–with more yield for the same or lower amount of risk and a an exposure to Treasury yields no worse than the ETFs.
Because of the apparent sensitivity of the correlation of portfolio returns to the 10-year Treasury yield, I have optimized the portfolio using five years of trailing data for the inputs to the risk analysis and the goal is to identify a portfolio which has the same trailing (5-year) risk as JXI with 0% correlation to the 10-year Treasury yield (matching JXI over this period) and the maximum possible yield. I have also specified a maximum of 10% allocation to any individual utility.
The resulting optimal yield portfolio is below:
|Empire Dist. Electric||EDE||10%|
|iShares Global Utilities||JXI||25%|
This portfolio has a 5.3% yield, 0% correlation to 10-year Treasury yield, and trailing five-year volatility equal to that of JXI. JXI has a 4% yield, as noted above. Over the last three years, this portfolio has a 2.6% correlation to 10-year Treasury yield and an annualized volatility of 10.5%.
The trailing three-year annualized total return is 4.7%.
This optimized portfolio of utility stocks and an ETF is a lot more competitive when compared to other high-income asset classes than IDU or JXI.
These results confirm what I suggested in my previous article: individual utility stocks may look attractive even when the broad index funds do not. Certain caveats apply. First, by holding concentrations in individual stocks, a portfolio is more exposed to the idiosyncratic risk of individual companies. In other words, moving from a broad ETF to concentrated holdings in individual stocks represents a shift in the type of risks in the portfolio if not the total portfolio risk. Taking on the idiosyncratic risk of an individual stock is a matter of investor preference.
An intriguing question that these results suggest is whether we might build a better utility ETF using the income-vs.-risk methodology. I have explored this question previously with bond index funds.