CFE 2015: Start Registration
View Submission - CFE
A0709
Topic: Contributions in panel data econometrics Title: Improving GMM efficiency in dynamic models for panel data with mean stationarity Authors:  Laura Magazzini - University of Verona (Italy)
Giorgio Calzolari - University of Firenze (Italy) [presenting]
Giorgio Calzolari - University of Firenze (Italy)
Abstract: Estimation of dynamic panel data models largely relies on GMM methods, and adopted sets of moment conditions exploit information up to the second moment of the variables. However, in many microeconomic applications the variables of interest are skewed, as for example in the analysis of individual wages, sizes of the firms, number of employees, etc.; therefore third moments might provide useful information for the estimation process. We propose a moment condition, to be added to the set of conditions customarily exploited in GMM estimation of dynamic panel data models, that is based on third moments. The moment condition we propose spans from the observation that, under mean stationarity, $y_{i0}$ can be written as the long run mean of the dependent variable and a randomly distributed error term. As a result, the individual effect $\alpha_i$ enters the first observation $y_{i0}$ multiplied by the same factor for each $i$ (equal to $1/(1-\beta)$). This data generating process for $y_{i0}$ is always adopted in the Monte Carlo simulations on dynamic panel data models that assume mean stationarity but not explicitly exploited. Monte Carlo experiments show remarkable efficiency improvements when the distribution of individual effects, and thus of $y_{i0}$, is indeed skewed.