Abstract
This note examines the use of indirect Monte Carlo methods in investigating the finite-sample properties of econometric estimators. It is found that the two-antithetic-variate method is far better than straight-forward simulation in estimating the biases of the estimators, but that a method combining the control-variate and the two-antithetic-variate techniques is better for estimating the dispersions. Comparisons with Nagar's approximations to the biases and dispersions are presented.

This publication has 0 references indexed in Scilit: