Abstract

RANGEL JIMENEZ, Andrés Eduardo. Relative superiority of the Kiviet and Blundell -Bond (GMM1) estimators in dynamic panels. A Monte Carlo experiment with finite samples. estud.gerenc. [online]. 2012, vol.28, n.125, pp.81-86. ISSN 0123-5923.

Given the widespread use of panel data in dynamic models, it is worth evaluating the performance of different estimators in finite samples in the presence of low and high persistence, with the latter being present in many macroeconomic series. This article analyzes the properties of the Least Square Dummy Variable (LSDV) estimators, Arellano -Bond Generalized Method of Moments Stage 1 (AB -GMM1), BBGMM1 (), AH (Anderson -Hsiao), and Kiviet using a Monte Carlo experiment. The results show that, in the presence of low persistence, the Kiviet estimator is the best performer based on the criteria of Root -Mean -Square Error (RMSE), bias and standard deviation. Meanwhile in the case of high persistence, the system estimator of Blundell and Bond (GMM1) is the best performing estimator against their rivals, followed by Kiviet estimator that exhibits good behavior, except in micropanels.

Keywords : Panel data; Dynamic models; Kiviet; Generalized method of moments.

        · abstract in Spanish | Portuguese     · text in Spanish     · Spanish ( pdf )