Mean-variance optimization is so well accepted that we often take it for gr
anted. In this article the authors examine the impact of relaxing its assum
ptions in a fixed-income context. They examine particularly whether the sub
stantial overweighting of non-Treasury bonds that is recommended in a mean-
variance context holds up under alternative return distributions and utilit
y functions. They also show how to incorporate scenario forecasting into an
optimization framework.