Bond Portfolio Strategies, Returns, and Skewness: A Note

The academic research has produced a series of contributions on optimal portfolio strategies (Bradley and Crane [1], Crane [4], Cheng [3], Fisher and Weil [5], Watson [9], Wolf [10]). Several of these studies--Bradley and Crane, Watson, and Wolf--conclude that an optimal strategy for bank portfolios would be a “dumbbell” strategy. Such a dumbbell strategy invests only in the shortest and longest maturities, ignoring the intermediate maturities. The logic is straightforward: liquidity risk is lowest in the shortest maturities and yield is generally highest in the longest maturities. The risk/return superiority of such a strategy was empirically verified by Watson, with subsequent confirmation by the Bradley and Crane tests via a stochastic dynamic programming formulation.