Date:         Sun, 16 Sep 2007 19:19:56 -0700
Reply-To:     Gerard <gerard9486@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Gerard <gerard9486@GMAIL.COM>
Organization: http://groups.google.com
Subject:      Seemingly Unrelated Regressions & Proc Model
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset="iso-8859-1"

Hi all

I am running seemingly unrelated regressions on the following two
equations:

EPSbfAbnAdj/BVperShare_t_1 = r + ((r-g1)/(1+g1))*((YendSP/q -
BVperShare)/BVperShare_t_1);            MEAN/BVperShare = g2 + ((r-g2)/
q)*(YendSP/BVperShare);

where r, g1, g2 and q are estimates.

As you can see, the parameter estimates of r and q need to be the same
across equations. I'm not sure if my code is correct. While it runs
fine but I am getting results that cannot be correct.

proc model data=&curr_file; *out=&curr_file._res itsur all;
        ods select all;
        parameters r1 r2 g1 g2 q;
        exogenous LHS_eq1 LHS_eq2 BVperShare_t_1 BVperShare YendSP; *YendSP
EPSbfAbnAdj BVperShare_t_1 MEAN BVperShare;

        EPSbfAbnAdj/BVperShare_t_1 = r1 + ((r1-g1)/(1+g1))*((YendSP/q -
BVperShare)/BVperShare_t_1);
        MEAN/BVperShare = g2 + ((r2-g2)/q)*(YendSP/BVperShare);

        outvars ASX_Code PriceToBook EPSbfAbnAdj BVperShare BVperShare_t_1
MEAN YendSP EPSRatio LHS_eq1 LHS_eq2 q r1 r2 g1 g2;

        fit / sur fsrsq normal itdetails prl=both outresid
out=&curr_file._out;

        run;
quit;

Any help will be appreciated.

Regards,
Gerard