12-20-2016 03:44 AM
I ran a cross-sectional regression (across 500 stocks) for a particular month. It is repeated for other 60 months (say yymonth is 201001-201512). Then, the average (across time) coefficients are calculated.
proc reg data=a outest=b tableout noprint;
model y = x1 x2 x3 / adjrsq ;
output out=RES r=rRESID;
Now I need to run proc syslin SUR grouping all the 60 months together.
Does anyone know how to 'choose' a month's data into each model equation? I think the code goes like this:
proc syslin data = a sur;
model1: model y = x1 x2 x3; <----this one include only 2010Jan
model2: model y = x1 x2 x3; <----this one include only 2010Feb
model60: model y = x1 x2 x3; <----this one include only 2015Dec
Thanks for your help!!
12-28-2016 09:41 AM
Wouldn't a BY statement be sufficient to accomplish this? In fact, the same BY statement that you used in PROC REG looks like it ought to work.