Moment Restrictions and Identification in Linear Dynamic Panel Data Models

Annals of Economics and Statistics

Tue Gorgens, Chirok Han, Sen Xue

Abstract

This paper investigates the relationship between moment restrictions and identification in simple linear AR(1) dynamic panel data models with fixed effects under standard minimal assumptions. The number of time periods is assumed to be small. The assumptions imply linear and quadratic moment restrictions which can be used for GMM estimation. The paper makes three points. First, contrary to common belief, the linear moment restrictions may fail to identify the autoregressive parameter even when it is known to be less than 1. Second, the quadratic moment restrictions provide full or partial identification in many of the cases where the linear moment restrictions do not. Third, the first moment restrictions can also be important for identification. Practical implications of the findings are illustrated using Monte Carlo simulations.

Keywords: Dynamic panel data models; Fixed effects; Identification; Generalized method of moments; Arellano-bond estimator

JEL classification: C230