We investigate the finite sample and asymptotic properties of
several estimation methods (Within-Group, GMM and LIML)
for a panel autoregressive structural equation model with random effects when both T and N are large.
When we use the forward-filtering to a structural model as Alvarez and Arellano (2003),
both the WG and GMM estimators are significantly biased when both T and N go to infinity
while T/N is different from zero. The LIML (limited information maximum likelihood) estimator has
consistency and the asymptotic normality when T/N converges to a constant as both T and N go to infinity.
Its asymptotic distribution has some bias and covariance which depend on the limiting behavior of T/N.
|