|Date: ||Sat, 5 Feb 2005 11:30:19 -0800|
|Sender: ||"SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>|
|From: ||"David L. Cassell" <cassell.david@EPAMAIL.EPA.GOV>|
|Subject: ||Re: PROC TSCSREG|
|Content-type: ||text/plain; charset=US-ASCII|
Jeffrey Newman <Jeffrey.Newman@BEA.GOV> wrote:
> I am using PROC TSCSREG to try to develop predicted metropolitan
> area income estimates. I'm aware that tscsreg does not compute
> observation-level statistics like residuals and predicted values.
> Have any of you had success in developing residual values based on
> the tscsreg procedure? My dependent variable is a log value of a
> growth rate (current year/previous year). Also, I want to use the
> Parks option in the model to account for autocorrelation in my
> pooled input data.
I haven't seen a way to get record-level information out of PROC
TSCSREG, not even using ODS. Sorry.
I would recommend that you try doing this model in PROC MIXED instead.
It doesn't do PARKS, but it will let you define a reasonable error
structure with an AR(1) or ARMA(1,1) model within each metropolitan
area. If you need more than that, you might want to consider trying
to run your estimates using PROC MODEL.
David Cassell, CSC
Senior computing specialist