When I began authoring my series on fixed income momentum strategies in January, I knew that my first strategy was going to involve high yield bonds and Treasury bonds. These two asset classes are inherently negatively correlated with high yield bonds rallying in good economic environments and selling off in bad economic environments where the probability of defaults by speculative grade companies rises accordingly. Treasuries typically have an inverse market relationship to high yield bonds, selling off in good economic environments due to a rise in inflation expectations that lowers real returns and rallying in poor economic environments as a flight-to-quality instrument.
While the Barclays Capital High Yield index made for an easy choice for the high yield leg of the momentum trade given its predominance in institutional indexing and its use as the basis for the largest high yield bond ETF (JNK), the choice of what Treasury benchmark to use was more difficult. I finally decided on simply the Barclays Treasury Index (replicated through the ETF GOVT) because it encapsulated the entire Treasury yield curve. Using the two broadest definitions of the two markets was the easiest way to demonstrate the efficacy of the momentum strategy.
Readers have since questioned why I did not use the Barclays Long Treasury Index (replicated through ETF TLT). Their point has merit. If the momentum strategy works in part due to the negative correlation of the two legs of the trade, adding a longer duration Treasury instrument provides more negative correlation and more upside in poor economic environments when the momentum strategy is likely to be in Treasuries.
Long Treasuries through the beginning of my series in January were capping off a thirty-year bull run that began with the Volcker-led disinflation of the early 1980s and was likely ending with the Bernanke-led quantitative easing of this cycle. My high yield bond index went back to 1983, when high yield bonds first came to the fixed income forefront due to their use in financing leveraged buyouts, and with a historically strong run for long Treasuries overlapping this dataset, I thought that this momentum strategy would offer too high of returns that could not possibly be replicated in forward periods because of the current level of interest rates.
Below are historical returns for high yield bonds, a rolling portfolio of Treasuries with greater than twenty years to maturity, and a momentum portfolio between the two. As readers of my other momentum articles have learned, the momentum portfolio owns the leg of the trade that had outperformed on a total return basis over the last month forward for the next one month before resetting.
Cumulative returns for the two legs of the trade are detailed below. The graph reads as follows - a dollar invested in the momentum portfolio in July 1983 would be worth $27 today.
There are some important takeaways from the table and graph above that I need to highlight:
1) Over the trailing 3, 5, 10, 20, and 30-year history of this trade, the momentum strategy has generated alpha, producing higher total returns than owning either high yield bonds or long Treasuries outright while exhibiting less return volatility than long Treasuries.
2) This outperformance has not held in the trailing twelve months, with this momentum strategy producing a negative total return. When Treasuries began their swoon in the early summer due to the pull forward of market expectations about the terminus of quantitative easing, both Treasuries and high yield bonds sold off as investors punted fixed income broadly. June 2013 marked only the second occurrence (October 2008) when both the long Treasury Index and the high yield bond index produced monthly returns worse than -2.5% in the same month in the last nineteen years. This type of positively correlated market move was exactly why I was hesitant to not use long Treasuries as a leg of the momentum trade at this point in time. The performance of the momentum strategy over the trailing twelve months has been its worst performance over that period since the interest rate selloff in 1994.
3) High-yield bonds exhibit less return volatility than long Treasuries. Some readers might find it interesting that "junk" bonds have lower risk by this measure than their longer duration Treasury counterparts. This is a function of the higher interest rate sensitivity of long Treasuries. I have demonstrated in past articles that long Treasuries currently have had higher trailing volatility than equities. If I am going to experience equity-like volatility, then I want average returns higher than 3.7%, the current yield on the thirty-year Treasury.
Guarantees are hard to come by in the investment world, but I believe that I can reasonably predict two things from this date:
1) The long Treasury/high yield momentum strategy will not replicate the 11.65% return that it generated over the trailing thirty years, but I would expect that over long-time intervals that the strategy will outperform high yield bonds and long Treasuries on a standalone basis.
2) The poor performance of the long Treasury/high yield bond momentum trade thus far in 2013 that made me hesitant to publish this strategy when I began this series will eventually normalize as the fixed income market finds its footing post-Fed tapering.
If readers would like to see this trade updated with market commentary in my monthly fixed income momentum strategy article, please respond in the comments section of this article. Look for my August versions of my three momentum monthly momentum articles over the next several days.