Bond Portfolio Optimization Using Dynamic Factor Models

Dynamic factor models for the yield curve have been extensively applied to fit and forecast the yield curve. We propose a novel utilization of these models in bond portfolio optimization. Specifically, we derive closed-form expressions for the vector of expected bond returns and for its covariance matrix based on a general class of dynamic factor models, and use these expressions to obtain optimal mean-variance bond portfolios. We also develop a duration-constrained, mean-variance optimization, which can be used to improve bond indexing. An empirical application involving two large data sets of U.S. Treasuries with different characteristics shows that the proposed portfolio policy outperforms a broad set of traditional yield curve strategies used in bond desks in terms of higher Sharpe ratios. Moreover, we find that an investor with a quadratic utility function is willing to pay a performance fee to adopt the proposed mean-variance bond portfolios. Finally, we discuss how an investor can benefit from adopting a dynamic rule to switch among alternative bond investment strategies. We find that the benefits of such dynamic portfolio selection rule are even more pronounced when the set of available policies is augmented with the proposed mean-variance portfolios