Delegated portfolio management and optimal allocation of portfolio managers

In this article, we investigate whether the application of the mean-variance framework on portfolio manager allocation offers any out-of-sample benefits compared to a naïve strategy of equal weighting. Based on an exclusive data-set of high-net-worth (HNW) investors, we utilize a wide variety of methodologies to estimate the input parameters including exponentially weighted moving average (EWMA), generalized autoregressive conditional heteroscedasticity (GARCH) and Bayes–Stein shrinkage estimation. We apply nine different mean-variance models, but find that none of these present any consistent benefit over a naïve strategy of equal weighting.