Duration-Adjusted Bond Fund Performance

Abstract

We uncover a previously neglected mechanical bias in bond fund performance due to the use of benchmarks with non-matching durations. We show that the duration bias is caused by the non-linear reaction of bonds with different durations to interest rate changes. We find empirically that the usual use of a broad bond index in previous research leads to a significant overestimation of average bond fund performance and to spurious findings of performance persistence. The key takeaway of our research is thus, that bond fund performance should be duration-adjusted by choosing for each fund the benchmark index which best matches its duration.