Are performance measures equally stable?

Managed portfolio performance evaluation is an important issue from both academics’ and investors’ points of view. One important aspect concerns the choice of the measure used to assess performance, because each index offers a different perspective about the trade-off between return level and risk exposure. A related crucial issue is the stability of measures over time and, hence, their predictive power about future results. In this work, we address the problem of evaluating of the stability of a performance measure. First, we discuss the use of alternative criteria to measure stability and propose a stability index based on changes of ranks over several periods of time. We also propose a statistical test to evaluate the homogeneity of the stability of alternative performance measures. Second, we suggest a composite performance measure, built as a linear combination of various indexes, specifically conceived to be maximally stable over time while preserving information about risk-adjusted return behavior. An application to a set of US mutual funds belonging to the Large Blend category shows how these methods work. Copyright Springer-Verlag 2012