Abstract
This study extends earlier results on bias-corrected estimators for the fixed-effects dynamic panel data model. We derive the inconsistency of the lsdv estimator for finite t and n large in case of both time-series and cross-section heteroscedasticity and show how to implement it in bias correction procedures.
Original language | English |
---|---|
Pages (from-to) | 220-227 |
Journal | Economics Letters |
Volume | 92 |
DOIs | |
Publication status | Published - 1 Jan 2006 |