| This thesis examines theoretical and empirical issues in evaluating the performance of bond portfolios, and develops a methodology to measure bond portfolio performance, using Long's numeraire portfolio. This measure defines the geometric difference between a portfolio's return and the return on a numeraire portfolio as the abnormal return of the portfolio.; Chapter 1 shows that the expected value of abnormal returns of a portfolio using the geometric difference can be interpreted as the net present value of investing in the portfolio and is the same as Jensen's alpha in the continuous-time limit. Unlike Jensen's alpha, this measure is robust to the existence of market timers, derivative securities or dynamic portfolio strategies. This chapter also suggests a way of constructing a numeraire portfolio in the bond market and examines its empirical properties. Also, this chapter suggests ways to reduce the variance of abnormal returns based on the numeraire portfolio approach to increase the power of the numeraire portfolio approach in performance evaluation.; Chapter 2 investigates the size and power of the numeraire-based performance measures as well as the size and power of Jensen's alpha, using simulations. The simulation results show: (1) For security pickers, both measures work well and the power of the numeraire-based measures seem to be comparable to that of Jensen's alpha; (2) For market timers, the numeraire-based performance measures have good power, while the power of Jensen's alpha depends heavily on Sharpe ratio; (3) Jensen's alpha doesn't value derivative assets correctly, while the numeraire-based measures do very well.; Chapter 3 examines the performance of government bond funds and corporate bond funds over the period of January 1976 to March 1995, using numeraire-based performance measures. The empirical findings in this chapter are: (1) On average, abnormal net returns on bond funds during the sample period are negative; (2) Expense ratios are negatively related to abnormal returns, and even after expenses are added back to returns, bond funds don't seem to outperform the market; (3) There is little evidence that the past performance of a bond fund predicts the future performance of the bond fund. |
