Introduction
The tools of modern portfolio theory1 are in general use in the equity markets, either in the form of portfolio optimization software or as an accepted framework in which the asset managers think about stock selection.2 In the ¯xed income market on the other hand, these tools seem irrelevant or inapplicable.Bond portfolios are nowadays mainly managed by a comparison of portfolio risk measures3 vis ¶a vis a benchmark.4 The portfolio manager's views about the future evolution of the term structure of interest rates translate themselves directly into a positioning relative to his benchmark, taking the risks of these deviations from the benchmark into account only in a very crude fashion, i.e. without really quantifying them probabilistically.5 This is quite surprising since sophisticated models for the evolution of interest rates are commonly used for interest rate derivatives pricing and the derivation of ¯xed
income risk measures.

Wilhelm (1992) explains the absence of modern portfolio tools in the ¯xed income markets with two factors:7 historically relatively stable interest rates and systematic di®erences between stocks and bonds that make an application of modern portfolio theory di±cult. These systematic di®erences relate mainly
to the ¯xed maturity of bonds. Whereas possible future stock prices become more dispersed as the time horizon widens, the bond price at maturity is ¯xed.8 This implies that the probabilistic models for stocks and bonds have to be di®erent and the tools of modern portfolio theory have to be adapted
to be applicable to ¯xed income instruments.