The overarching question in the management of domestic bond portfolios is the future direction of interest rates. The slow economic recovery in the United States and the resultant accompanying accommodative monetary policy, coupled with the flight to quality bid for high quality assets driven by the economic uncertainty emanating from Europe, has pushed rates on U.S. Treasury securities to historic lows. Falling interest rates always cause a paradoxical situation for bond investors as lower rates increase the value of their bond portfolios, but lower the returns earned on future re-investment of coupon. Conventional wisdom posits that Treasury bond investors will refuse to accept nominal rates lower than long-run inflation indefinitely, and rates will rise accordingly to compensate. Correctly timing this rate move is imperative to capturing value for your portfolio.
How do investors think about timing rising interest rates in order to fend off the negative effects on the value of their bond portfolio? Active bond investors must believe that their implementation of tactical changes in the maturity profile or credit quality can produce outsized gains relative to the market. This article presents two simple to implement trading strategies to generate higher levels of risk-adjusted returns from your bond portfolio: one provided by a recent academic paper by Naomi Boyd of West Virginia University and Jeffery Mercer of Texas Tech University and one provided by the article's author, Seeking Alpha's Ploutos.
In their paper "Gains From Active Bond Portfolio Management Strategies," the aforementioned authors studied the value of simple bond management strategies that reallocate portfolio holdings on the basis of identified inflection points in the interest rate cycle. The authors utilized a rolling Markowitz (mean-variance) optimization approach that relies only on past information to trigger re-allocations, allowing portfolio managers to use these techniques to drive real-time changes in allocations.
For individual investors or portfolio managers seeking alpha from the rate component within a fixed income portfolio, one may manage the duration of the portfolio by shifting allocations between long and short positions in the maturity profile. Duration carries a dual meaning in fixed income portfolio management. Duration is a measure of the interest rate sensitivity of the portfolio to parallel shifts in the yield curve, and also the weighted average timing of expected future cash flows from the bond holdings. The academics examine the effectiveness a common trading strategy - rate anticipation trades - that uses past information to inform the decision to shorten or extend duration in a bond portfolio.
Since bond prices move inversely to the level of interest rates, bond investors using the rate anticipation strategy will extend the duration of their portfolio to capture capital gains when they predict interest rates will decline; alternatively, they will shorten their portfolio duration to fend off market value declines if they forecast an increase in interest rates. The authors investigate whether portfolio performance can be improved by shifting portfolio allocations on the basis of shifts in the interest rate cycle. The authors define an interest rate cycle by changes in the discount rate set by the Federal Open Market Committee (FOMC), noting that changes in the direction of the discount rate are rare (14 in the thirty-six year study period). Falling interest rate phases begin when the discount rate is decreased by the FOMC and end when the discount rate is increased; the opposite is true for rising interest rate phases.
In their study, the authors created long-only Markowitz-efficient portfolios to test whether the rate cycle strategy would have performed well in practice. To determine optimal real-time re-allocations, the authors use a rolling optimization approach where the performance of the past rate cycle is used to form optimal portfolio weights at three different risk levels. These risk tolerances set caps on the amount of volatility the portfolio can experience. (For example, the first interest rate cycle was a period of twenty-three months. The authors used this data to form optimal portfolio weights of different bond indices that would cap variance of returns at pre-defined levels. These weights were applied over the next interest rate cycle of 33 months. The authors then used the data for the first 56 months to form their portfolio weights for the next interest rate cycle, marked by a directional change in the discount rate.) The rolling optimization approach outperform the Lehman Brothers (now Barclays) U.S. Aggregate Index and the Lehman Brothers U.S. Credit Index, two widely used fixed-income benchmarks, based on annual returns and standard deviations and on monthly return-to-risk ratios for the respective periods. Even the most aggressively constructed portfolio outperformed the Barclays Aggregate Index while still demonstrating less volatile returns, producing a 10.58% annual return with a 5.25% annual standard deviation to the index's 8.67% return and 6.07% annual standard deviation.
The authors next investigate several rate anticipation strategies by using only two bond categories and re-allocations according to an extreme weighting strategy of either 0 or 100 percent at the cycle turning points. For the rate anticipation strategy, the authors compare (1) long-term Treasury bond with Treasury bills. For the inter-market spread strategies, the authors compare (2) high yield securities with long-term government securities and (3) long-term corporate bonds with long-term government securities. Two category allocations blend rate anticipation and intermarket spread strategy characteristics: (4) high yield bonds and Treasury bills and (5) long duration corporate bonds and Treasury bills. The only pair of note to me was a strategy that invested in T-bills in a rising rate environment and high yield in a falling rate environment, which produced an annualized return of 10.91% and an annual standard deviation of 5.83%, producing a higher return than the index with less variance of returns. This strategy would be easy to replicate for Seeking Alpha investors, and would mean that investors would continue to hold high yield securities given that we have not reached an interest rate cycle inflection point, but were likely to be invested in Treasury bills in the intermediate term when the Fed finally begins to increase interest rates in 2014 or 2015.
While my bond portfolio management and trading guide differs from those constructed by Professors Boyd and Mercer, it features a common principle - letting past performance guide future performance. I have frequently written about momentum in fixed income portfolio performance in past articles. "Momentum and High Yield Bonds" demonstrated a switching process between relative ratings cohorts based on trailing performance that generated positive out-performance in subsequent periods. "Improving Your Return Profile with a Simple Momentum Strategy" demonstrated that investors who rotated between a standard fixed income benchmark and equity benchmark based on trailing period performance could also enhance their return profile while significantly reducing the volatility of their portfolio. It should then be no surprise to readers that momentum instructs my bond portfolio construction.
The most oft replicated corporate bond benchmark is the Barclays (formerly Lehman Brothers) U.S. Corporate Index. For my study, I examined the intermediate segment (duration 4.6 years) and the long segment (duration 13.6 years) of the investment grade portion of this index using data from January 1973 through the present. My study documents a periodic structure of average returns to portfolio strategies that purchased the corporate bond maturity bucket (intermediate or long) that had produced a higher total return in trailing periods, and holds that bucket for a matched duration forward period. For the study, the lag period of historical performance was set equal to the forward holding period (i.e. if we based the formulation of the momentum portfolio on whether the intermediate segment or long segment had outperformed in the trailing month, we would hold the higher returning segment for the next one month.)
The table below demonstrates that a momentum portfolio that held the segment of the bond portfolio that had outperformed over the past one month forward for one additional month, produced higher absolute returns. This momentum portfolio also produced its higher absolute return profile with only 81% of the variance of the long segment of the corporate bond universe. A momentum strategy that switches between the intermediate and long part of the curve dependent on past relative performance outperforms a buy-and-hold strategy of long credit.
The table below demonstrates that a momentum portfolio that held the segment of the bond portfolio that had outperformed over the past one quarter forward for one additional quarter, produced higher risk-adjusted returns than owning solely the long portion of the corporate bond universe. This momentum portfolio also produced its higher absolute return profile with only 80% of the variance of the long segment of the corporate bond universe. The momentum factor dissipated slightly with the quarterly holding period, and no longer outperformed on an absolute basis. (Note: the slight differences in the annualized averages between the monthly and quarterly methodologies are due to the small differences in the study horizon, which began in February 1973 for the monthly analysis and second quarter 1973 for the quarterly analysis in order to encapsulate a trailing period.)
Momentum strategies with matched lag and forward holding periods of six months and one year respectively also provide better risk-adjusted returns than a buy-and-hold strategy in the long end of the curve. Astute observers will notice that the intermediate part of the curve has produced higher risk adjusted returns over long-time periods than the long-end of the curve, please look for an upcoming article that examines how to profitably exploit this anomaly.
Past literature and results from both the Boyd/Mercer study and my own analysis indicate that risk-adjusted returns on fixed income portfolio could have been substantially improved by using only easily observable past information to reallocate portfolio weights. From my analysis, bond investors who are seeking additional yield by moving to the thirty-year part of the curve should look to shorten duration in the month after the intermediate portion of the curve outperforms the long end. With the long end outperforming in both April and May, it may be prudent to continue a long duration stance, and see if the modest selloff in the long end of the yield curve in June continues.
I hope this analysis helps readers think about their bond portfolio allocation by offering easy to implement trading cues to guide their portfolio construction and tips on how to re-allocate their bond portfolio to generate market-beating risk-adjusted returns given the eventuality of a rising rate environment.