On the relative importance of duration constraints

This paper uncovers an implicit assumption, and its implications, made in the process of maximizing yield (or minimizing costs) subject to the duration constraints. Using linear programming results, it is shown that this technique is sensible only if the yield of a bond is a linear function of its duration measures. Utilizing this result the paper analyzes the relative importance of the duration constraints. Former studies have hinted that the first order duration may be the most important. It is shown here there is no reason to satisfy the first duration constraint with priority over satisfaction of a higher order duration constraint. Practitioners who use duration techniques for portfolio immunization must be aware of such an important counter-intuitive result.