PROC MODEL for Multiple Models

Junyong — Thu, 20 Feb 2020 19:30:58 GMT

Suppose there is a regression.

data have;
do t=1 to 5000;
x=rannor(1);
w=rannor(1);
y=x+w+rannor(1);
output;
end;
run;

PROC REG easily estimates the following three regressions.

proc reg;
model y=x;
model y=w;
model y=x w;
quit;

But I need to use a Newey–West standard error, i.e. PROC MODEL. I tried the following, which didn't work correctly.

proc model;
y=a+b*x;
fit y/gmm kernel=(bart,2,0) vardef=n;
y=c+d*w;
fit y/gmm kernel=(bart,2,0) vardef=n;
y=e+f*x+g*w;
fit y/gmm kernel=(bart,2,0) vardef=n;
quit;

Should one repeat PROC MODEL thrice to estimate the three regressions?

Re: PROC MODEL for Multiple Models

ChrisNZ — Fri, 21 Feb 2020 03:16:59 GMT

Moved to Procedures forum.

topic Re: PROC MODEL for Multiple Models in SAS Procedures

PROC MODEL for Multiple Models

Re: PROC MODEL for Multiple Models