Please use this identifier to cite or link to this item: http://hdl.handle.net/10419/119532
Authors: 
Biørn, Erik
Han, Xuehui
Year of Publication: 
2015
Series/Report no.: 
Memorandum, Department of Economics, University of Oslo 4/2015
Abstract: 
GMM estimation of autoregressive panel data equations in error-ridden variables when the noise has memory, is considered. The impact of variation in the memory length in signal and noise spread and in the degree of individual heterogeneity are discussed with respect to finite sample bias, using Monte Carlo simulations. Also explored are also the impact of the strength of autocorrelation and the size of the IV set. GMM procedures using IVs in differences on equations in levels, in general perform better in small samples than procedures using IVs in levels on equations in differences. A case-study of the impact of Foreign Direct Investment (FDI) on GDP, inter alia, contrasting the manufacturing and the service sector, based on country panel data supplements the simulation results.
Subjects: 
Panel data
Measurement error
ARMA
GMM
Error memory
Monte Carlo
Foreign Direct Investment
Economic development
Country panel
JEL: 
C21
C23
C31
C33
O11
O14
Document Type: 
Working Paper

Files in This Item:
File
Size
304.81 kB





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.