The Effect of Errors in Means, Variances, and Covariances on Optimal Portfolio Choice

Statistical inference procedure for the mean–variance efficient frontier with estimated parameters

“…At the 5% level of significance, two changes are detected by T S and only one byT S . This result is in line with the finding of Merton (1980), Best and Grauer (1991), Chopra and Ziemba (1993), who showed that impact of the estimation error on the means of the asset returns is much larger than the impact of errors on the variances and covariances. Consequently, the distribution of test statistics that depend on the sample mean can be significantly influenced by the estimation error in the means and, thus, cannot detect small changes of the corresponding parameters.…”

Optimization, cointegration and diversification gains from international portfolios: an out-of-sample analysis

“…Thus, the number of assets with non-zero and positive weights in the optimized portfolio will typically increase as the adjustments help balance the extremely high and low average returns on various assets. This reduces the estimation error and contributes to the improvement in out-of-sample performance of the historically optimized portfolio (e.g., Chopra and Ziemba 1993).…”

Asymptotic Multivariate Dominance: A Financial Application

“…For this reason, we propose a modification to the meanvariance approach presented in the previous example by taking into account the effect that estimation error may have on the determination of the efficient frontiers. Typically, we are concerned with the error in the estimation of the mean, because of its major impact on portfolio choices (Chopra and Ziemba 1993). In the financial literature, many approaches of robust portfolio optimization have been proposed in order to ensure that decisions are adequate even if estimates of the input parameters are incorrect.…”