Empirical Research On Bond Portfolio Optimization

This study proposes the use of modern portfolio theory to select optimized portfolios of People's Bank Of China (PBOC) bonds. The goal of this study is to characterize the risk-return profiles that can be achieved, both in-sample and out-sample. Despite its broad acceptance in the equity markets, the mean-variance framework has never quite established a foothold in the fixed-income markets. Difficulties to adapt the mean-variance framework for bond portfolios might explain it has not been much used by bond market practitioners.This paper adapts the bond portfolio selection model proposed by Wilhelm to portfolios of coupon-bearing bonds. Based on the 1-factor and 2-factor Vasicek models, it derives the expected values, variances and covariances of bond returns that are needed to solve the portfolio optimization problem. Using data of the interbank PBOC bond market, it then estimates the parameters of the term structure models. Both 1-factor and 2-factor models are calibrated by using the Kalman filter technique. Based on the resulting estimates, this study investigates the risk-return profiles of optimized portfolios. Finally, it contrasts these predictions with the realized risk-return profiles.The main result of this paper is that optimized bond portfolios exhibit quite attractive risk-return profiles.