From: Selina on
Hi,

I need to calculate Newey west autocorrelation-adjusted standard error
for a single time series variable. I have found some codes for the
newey west standard errors for regressions. (http://
www.listserv.uga.edu/cgi-bin/wa?A2=ind0306d&L=sas-l&P=9888). However,
I am not using this for regression, but for a single variable only.

I tried to regress the variable on a constant, however, the output of
"parameterestimates" doesn't seem to be the standard errors that I
need. I have tried all the output files that this "Proc Model" with
GMM option could produce, but still have no idea where to get the
right NW standard errors. My codes are below. Please help me to see
if there is anything wrong with it. Thank you so much!

==================================================
%let lags = 2;
ods output parameterestimates = x_nw;
ods trace on;
proc model data= input ;
by firm_id;
parms a;
x= a ;
instruments / intonly;
fit x/ gmm kernel=(bart,%eval(&lags+1),0);
run;
quit;
ods trace off;
ods output close;
====================================================

Regards,
Shuang