Yield Curve Attribution for Global Bond Portfolios

Performance attribution is critical for every type of investment management style — both discretionary and systematic — and is important for both portfolio managers and investors. The concerns that managers and investors have are fourfold: (1) assessing portfolio manager skill, (2) measuring returns, (3) identifying the sources of risk, and (4) identifying the sources of drawdowns or losses. These four concerns are related to managers’ and investors’ concern with their portfolios outperforming their respective benchmarks. The focus of this chapter is to describe a yield curve-based approach to performance attribution for global bond portfolios.We begin by discussing an approach to performance attribution that decomposes all sources of bond portfolio return and then adds the impact of foreign exchange exposure to the total sum of bond returns in local currencies. An important advantage of this approach over return-based regression-based models is that no assumption of model composition, parameters, and explanatory factors is made. Furthermore, the model is nonlinear and thus faithfully captures the dynamics of bond prices. Second, we show how a bond portfolio's return components (summarized as bond and currency effects) behave depending on the base currency of the fund. Third, we discuss the interaction between pricing and data quality, which is related to model estimation and the estimation of errors in performance attribution analysis. Whereas pricing and data quality might be considered exogenous factors, model specification, model estimation, and model errors are fundamental determinants of the outcome of performance attribution analysis.