Estimation and inference in dynamic unbalanced

Transcript

Estimation and inference in dynamic unbalanced
The Stata Journal (2005)
5, Number 4, pp. 473–500
Estimation and inference in dynamic
unbalanced panel-data models with a small
number of individuals
Giovanni S. F. Bruno
Istituto di Economia Politica, Bocconi University, Milan
Abstract. This article describes a new Stata routine, xtlsdvc, that computes
bias-corrected least-squares dummy variable (LSDV) estimators and their bootstrap variance–covariance matrix for dynamic (possibly) unbalanced panel-data
models with strictly exogenous regressors. A Monte Carlo analysis is carried out
to evaluate the finite-sample performance of the bias-corrected LSDV estimators in
comparison to the original LSDV estimator and three popular N -consistent estimators: Arellano–Bond, Anderson–Hsiao and Blundell–Bond. Results strongly support the bias-corrected LSDV estimators according to bias and root mean squared
error criteria when the number of individuals is small.
Keywords: st0091, xtlsdvc, bias approximation, unbalanced panels, dynamic panel
data, LSDV estimator, Monte Carlo experiment, bootstrap variance–covariance
c 2005 StataCorp LP
st0091