Mean-variance efficient portfolios are influenced by errors due to approximation, estimation and incomplete information. Therefore, the obtained results - recommendations for the risk and portfolio manager, should be carefully analyzed.
This note presents results of a simulation study devoted to the output analysis with respect to perturbed input data - expected yields and elements of their covariance matrix. The motivation comes from results of the simulation study of Chopra and Ziemba [1993], whose conclusions about the prevailing importance of expectations turn out to be substantially influenced by the chosen value of the model parameter that quantifies the level of the risk aversion of the investor.
Our simulation study complements these results comparing the influence of perturbed values of expectations, variances and covariances of yields for the whole range of the risk aversion parameter.