Date: Tue, 18 May 1999 15:15:51 -0700
Reply-To: "ZHANG, XIAOBO" <X.ZHANG@CGIAR.ORG>
Sender: "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From: "ZHANG, XIAOBO" <X.ZHANG@CGIAR.ORG>
Subject: adjustment for spatial covariance matrix
I am trying to run a regression using a panel data set (N>>T), in which I
found there are strong spatial correlations. It seems that the covariance
estimation in the ETS/TSCSREG procedure is invalid because the N is much
greater than T. Is there anyone knowing how to adjust the spatial covariance
matrix in a cross-section or panel analysis? Also, I would like to know
whether a dynamic panel analysis can be conducted in SAS or not.
Any macro/program code, or advice are greatly appreciated.