Government bond markets experienced sharp moves in 2018. As a result, investors could add value through bond market timing, gaining protection from rising yields and benefiting from declining ones. In this context, the positioning of our Dynamic Duration funds was indeed truly dynamic.
Active duration management once again proved to be an effective way to avoid human biases and benefit from out-of-consensus positions. Systematically sticking to proven bond market drivers without any bias allowed the fund to embrace government bonds and take a maximum overweight in the last months of the year. With strong returns in this difficult period for risky assets, the two funds delivered valuable diversification just when investors needed it most.
Last year, the Robeco QI Global Dynamic Duration and QI Long/Short Dynamic Duration generated positive total returns of 1.7% and 1.6% respectively, gross of fees, significantly outperforming their respective benchmarks.1
The duration positions taken in the Dynamic Duration funds are based on the outcome of our duration model. This model was developed in the early 1990s and uses six variables to forecast the direction of bond yield movements: three fundamental variables relating to macroeconomic drivers of bond markets such as economic growth, inflation and monetary policy; one valuation variable to assess to what extent these expectations are already discounted in bond prices and two technical variables (trend and season) to improve the timing of the model.
Four out of the six variables contributed to the strong relative performance in 2018, while the other two made a neutral contribution. The economic growth and inflation variables contributed positively, as they signaled both the rise in yields at the start of the year and the bond market rally at the end. The inflation variable gave up part of its performance in the Italian turmoil, while the growth variable had a weaker period in the summer. The monetary policy and valuation variables contributed quite consistently for most of the year by signaling that US bonds would underperform. The trend variable and the seasonal variable detracted from performance in the first part of the year but recovered in the rally in the last months of the year.
The enhanced trend variable would have generated a slightly better result over the full research period
As part of our continuous efforts to optimize the duration model, our researchers recently analyzed its trend variable. The trend variable is combined with a seasonal variable to optimize timing in the model. Specifically, our researchers analyzed the consequences of replacing the existing trend variable with a somewhat ‘longer-term’ one. They also looked at ways to improve the variable design to avoid short-term mean-reversion effects.
They found that the enhanced trend variable would have generated a slightly better result over the full research period, due to better performance in the last two decades. Meanwhile, the turnover of the strategy would have been consistently lower during the research period, thanks to this ‘longer-term’ trend variable, in particular in the ten years from 2008 to 2018. As a result, Robeco recently decided to enhance its duration model with this new trend variable.
1 Source: Robeco Performance Measurement. All figures in EUR, DH share, gross of fees. In reality costs (such as management fees and other costs) are charged. These have a negative effect on the returns shown. Past results are no guarantee for the future.