The distributions of the Maximum Empirical Likelihood (MEL) estimator and
the Generalized Method of Moments (GMM) estimator for the coefficient of one endogenous
variable in a linear structural equation are evaluated numerically. Tables
and figures are given for enough values of the parameters to cover most of interest.
Comparisons of the distributions of the MEL estimator and the GMM estimator
with their asymptotic expansions are made. We find that the MEL estimator does
not have any moments of positive integer orders and careful analyses are needed
to use its bias and mean squared error because usually they do not exist.
|