June 25, 2009, 12:45–14:00
Toulouse
Room MF 323
Brown Bag Seminar
Abstract
This paper tries to fill a gap in the existing literature on parametric estimation of production functions by extending GMM methods to unbalanced panel data. Using a panel data set of Chilean manufacturing firms over a 15 years period, this paper compares the performance of the Difference and System GMM (Arellano and Bond [1991], Blundell and Bond [1998]) with the widely popular method of Levinsohn and Petrin (2003). It conducts a sector based analysis similar to the one performed in Levisohn and Petrin (2003) by applying bootstrapping techniques to every subsample. Our work shows that GMM estimation allows for greater flexibility while it seems to deliver much robust results.