Selection of control variates for variance reduction in a multiresponse simulation with a small number of replications

Abstract

In this paper the selection of control variates for simulation experiments with multiple response variables is discussed. The aim is to reduce the variance of all response variables as much as possible. Therefore, response variables are weighted according to their coefficient of variation. A criterion is developed to determine the number of control variates and select the most appropriate ones. When the number of replications is small due to computational costs, outliers may seriously effect the estimation of coefficients for the control variates. Robust regression, based on the least median of squares criterion, is used to detect outliers.