Prominent investors, including Bill Gross and Warren Buffett, now say that the yields on long-term government debt do not justify the risks. But is this perception correct? I offer a way to answer that question – and to construct a low-risk high-income portfolio – using the prices of put options to derive the true risk levels of various asset classes.
We have seen some extraordinary pronouncements with regard to U.S. government bonds this spring. Both Gross and Buffett have said that government debt, particularly long-term government debt, is not providing sufficient yield to be worth the risks.
There are many fundamental reasons why one might conclude that government bonds, particularly those with longer duration, are not an attractive proposition. First, there is inflation risk. There are also the enormous long-term deficits that the U.S. is facing. Neither PIMCO nor Buffett suggests that the U.S. is likely to default, but rather that there is a high level of risk of reduced real purchasing power associated with owning dollar-denominated bonds.
The low yields on U.S. bonds, however, say the capital markets do not agree. Investors apparently feel that they are being fairly compensated for the risks that they face with long-term government bonds. While yields on 30-year Treasury bonds are higher than they were back in September and October of 2010, they have fallen from 4.7% earlier this year to 4.3% in mid-May. For investors to be taking on the risk of lending money to the U.S. government at a yield of 4.3% (or even 4.7% for that matter) for 30 years, they must feel that their risk is fairly low.
Let’s look at how the various risks associated with debt instruments can be calibrated and how that information can guide your portfolio decisions.
Measuring income risk
In previous articles, I have proposed a simple and effective approach to quantifying the risks of holding income-generating assets. With the advent of ETFs and options traded on ETFs, we have a valuable source of information on the market’s consensus view of the risks associated with an investment. Investors can buy put options, for example, on the iShares 20+ year government bond ETF (TLT). These put options put a calculable price on the risks associated with owning long-term bonds. The measure of risk derived from options is called implied volatility, the risk level that explains the current prices of options. The higher the implied volatility, the more expensive the option will be. A variety of sources report implied volatility, including Morningstar.
Implied volatility is a measure of total risk: It represents that price at which we can obtain insurance against a decline in the value of a specific stock, bond or ETF. For a bond ETF, the principal risks are default and depreciation of the underlying currency. For high-yield bond ETFs (such as HYG or JNK), the risk is primarily from default. For the U.S. government, we are not concerned with default, but we are concerned that the purchasing power of dollars may decline. For an individual stock, the principal risks are a combination of market risk and company-specific risk. For master limited partnerships (MLPs), the principal risk is a reduction in income. For an income investor, the availability of an aggregate risk measure in the form of implied volatility on ETFs is of enormous value.
Why, you may ask, is implied volatility is a better predictor of future risk than simply looking at historical volatility? There is an extensive body of research devoted to this question. While much of the literature has focused on volatility of market indexes (such as the S&P500), there is also evidence that implied volatility for individual stocks provides a useful predictive measure, and I have found this to be the case in my own research. Furthermore, option-implied volatility correlates with credit default spreads (CDS). The CDS market is generally consistent with the options market in assessing the risk associated with a given company.
Investors think of bond yield as a measure of risk. Junk bonds offer higher yields than investment-grade bonds, and longer-duration government bonds offer higher yields than shorter-duration bonds. Yield is not truly a measure of risk, however. It is not hard to see why using bond yield as a risk outlook is problematic. If investors become especially fearful of equity risk, for example, they put their money in bonds. As money flows into bonds, bond yields drop. It would be irrational to suggest that government bonds are less risky after investors pour huge amounts of money into this asset class. Granted, long-term bonds still have higher yields than short-term bonds, but the crucial question is how risky bond yields look when compared to other asset classes.
Are the yields on long-term government bonds high enough to motivate income investors to buy these bonds, as compared to other income-generating alternatives? PIMCO and Buffett suggest that the risk is too high relative to the yield. By comparing yield to implied volatilities, we can examine this question from the perspective of the actual market prices of ‘insuring’ these risks.
In an article published on October 1, 2010, I examined the yield versus risk of a range of asset classes. I concluded that “the really striking result is that high-yield corporate bonds have dramatically higher yields than long-term government bonds, with about the same level of risk. Similarly, investment-grade corporate bonds provide more yield with less risk than intermediate-term government bonds.” On this basis, corporate bonds looked more attractive than government bonds.
How have long-term bonds performed in the interim? Over the seven and a half months since I published my analysis, the market has taken a consensus view that supports my results. Long-term government bonds have substantially declined, while high-yield bonds have rallied (see chart below).
To move beyond using the yield-vs.-risk measure for individual asset classes, I proposed a version of the so-called “efficient frontier.” In the traditional efficient frontier, risk (measured by volatility) is plotted on the horizontal axis and expected return is plotted on the vertical axis. Calculations can then identify portfolios with the maximum expected return at each risk level. I proposed creating portfolios with the maximum yield for a given level of risk.
I explored this approach in my work in October and also more recently in examining the probability that an individual company will cut its dividend. To perform this calculation, we need to combine outlooks for risk with correlations between individual assets. I did this using my Monte Carlo portfolio analysis tool (Quantext Portfolio Planner). The outlooks for risk are very similar to the implied volatilities, but they are not identical because of the requirements of maintaining the correlations between assets.
The fact that long-term government bonds looked unattractive on a standalone basis back in October did not suggest that there was no place for an investment in this asset class. When I attempted to identify a portfolio that provided the highest yield for a given level of risk, I identified a ‘fair value’ income portfolio (one that was estimated to be trading neither above nor below is fair price) with a yield of 6.8% and half the expected volatility of the S&P500. This portfolio had a 15% allocation to a long-term government bond fund (TLT).
Current results: Outlook as of May 2011
I have updated the calculation of the yield-risk frontier using data available through April 2011 (below) in order to show how the optimal yield portfolios have changed. I added a constraint that no ETF could have an allocation greater than 35% of the total portfolio. The results below show the Monte Carlo simulations for select individual stocks and ETFs, as well as the optimal portfolios.
The optimal portfolios were created using a selection of individual stocks, MLPs, REITs, and ETFs, limiting the list of possible investments to those with the highest current yield-to-risk ratios. Short-, intermediate-, and long-term government bond ETFs made the cut, as did high-yield bonds and investment-grade bond ETFs. Telecoms remain attractive, too – especially international telecoms. MLPs are playing a larger role in optimal yield portfolios than in my original analysis.
The projected volatilities for the individual funds match their option-implied volatilities very closely. The largest difference between implied volatility and projected volatility is for high-yield bonds (HYG). The model projects the volatility to be 19.2% vs. 14% for the longest dated put options (expiring in December 2011). As of this writing, long-dated at-the-money put options on the S&P500 (SPY) have implied volatility of about 21% (Note: At-the-money options are those with strike prices very close to the current price of the underlying asset – in this case, the price of HYG).
Risk-Yield Frontier (data through April 2011)
The long-term government bond ETF (TLT) has expected volatility of 17.8%. The at-the-money implied volatility for this ETF as of May 18 is 17.4% for options expiring in Jan 2012 and about 24% for options expiring in Jan 2013. For the risk level associated with this index of long-term bonds, the yield of 5.1% is not attractive.
There are many possible portfolios with less risk and substantially higher yield than long-term bonds – as evidenced by how far the optimal curve is above TLT. The most substantial practical consideration in building a portfolio near this frontier is the maximum allowable allocation to a single stock. One candidate portfolio is:
7.4% Yield / 17.2% Volatility Portfolio
This portfolio is less risky than the S&P500 (which has implied volatility of 21%) and has substantial allocations to MLPs and telecom firms. The largest single component of this portfolio is high-yield bonds. Government bonds play a very small role in this portfolio, with a 3% allocation. There are portfolios near the optimal frontier that have about the same risk and higher allocations to long-term bonds, but none that I have identified has an allocation to TLT above 5%.
By accounting for portfolio risk and yield in a portfolio analysis that included the effects of the correlations between assets, we can determine the maximum yield available for a given amount of risk. The results from this analysis indicate that long-term government bonds are far below the efficient frontier. No single asset will necessarily be close to the frontier, because we expect that any optimal portfolio will combine asset classes, but the optimal portfolio has a small allocation to long-term government bonds and a substantial allocation to high-yield bonds for portfolios with risk levels generally comparable to those of long-term bonds.
One of the questions that came up in response to my earlier research on yield vs. risk was the role of price appreciation: Should we also control for total expected return? The answer is yes. Total return is the sum of income and price appreciation and is ultimately what matters to investors. There are a variety of ways to estimate total expected return. The Monte Carlo simulation (QPP) generates expected total returns for individual asset classes and portfolios. The expected annual long-term total return for the model portfolio above is 8.5% and the expected total return for TLT is 6.8%. These are long-term asset class-based return projections. There is little question, however, that the expected price appreciation is considerably less certain than future yield, regardless of method.
It is possible to create a portfolio with a higher yield, much less risk, and about the same expected total return as TLT:
5.7% Yield / 12% Volatility Portfolio
This portfolio has expected total return of 6.6% (versus 6.8% for TLT), and the risk on this portfolio is 33% less than TLT (12% volatility vs. 17.8% for TLT). This portfolio accomplishes this with only 7.5% allocated to TLT, plus an additional allocation to shorter-term government bonds.
Whether your goal is to get the highest yield available with about the same risk as long-term bonds, or whether it is to receive a yield comparable to long-term bonds for the lowest possible risk, long-term government bonds will play a small part in the optimal portfolio.
Geoff Considine is the founder of Quantext, Inc. and is the developer of Quantext Portfolio Planner (QPP), a portfolio management tool for advisors. More information is available at www.quantext.com.
Quantext is a strategic adviser to FOLIOfn,Inc. (www.foliofn.com), an innovative brokerage firm specializing in offering and trading portfolios for advisors and individual investors.
Read more articles by Geoff Considine