Performance of Asset Allocation Strategies in Europe: An Out-of-Sample Assessment

Abstract

Optimal diversification gain is frequently more than offset by estimation error, leading to a well documented gap between the in-sample and out-of-sample performance of investment strategies. Our findings suggest that popular attempts to address the estimation error can be counterproductive in improving the realized performance. In particular, potential diversification deteriorates if the original strategy outperforms the naïve benchmark; whereas the performance improves – up to the suboptimal 1/N return levels – in the case of bad performing strategies. We argue that constraining the weight variability reduces the negative effects of the estimation error but they still fail to improve the diversification potential.

Keywords: asset allocation, portfolio choice, estimation error, out-of-sample performance, European portfolios

JEL Classification: C58, G11, G15