Minimising operational risk in portfolio allocation decisions

Abstract

Models play an important role in strategic asset allocation (SAA). Overdependence on the stability of models, however, can reduce the utility of the results for practitioners. Professionals face the problem of choosing an SAA model that matches their goals, and must bear the operational risk of making the wrong choice (the model risk). The present paper sets out to assist in this challenge by evaluating several methodologies for estimating efficient portfolios, according to the perspective of a global long-term investor. It also presents the resampling adjusted technique (RATE) approach to incorporate estimation risk into mean-variance portfolio selection, based on the portfolio resampling technique. The results support the use of the Michaud resampling methodology followed by the RATE, as they offer better results in terms of financial efficiency, allocation stability and diversification. The evaluation of the different models uses several international asset classes for the period between June 1998 and July 2006. The findings are very useful for practitioners who can benefit from a fairly simple and robust asset allocation methodology.