TSD-consistent performance assessment of mutual funds

This paper presents three new data envelopment analysis–based approaches to assess the relative efficiency of mutual funds (MFs). Each model considers an appropriate risk measure as input and an appropriate return measure as output. The risk and return measures have been chosen so that the proposed models are consistent with third–order stochastic dominance (TSD) rules. This means that the MFs found efficient by the proposed models are also, in a necessity condition sense, TSD efficient and therefore of highest consideration for all non–satiated, risk averse investors that also have decreasing absolute risk aversion. The proposed approach is illustrated with real data on a set of Spanish MFs and compared with existing approaches from the literature based on Mean–Variance and Mean–Variance–Skewness models.