How to get Newey West standard errors for large a dataset/regression?

somebody — Sun, 13 Jun 2021 04:10:03 GMT

I want to get the Newey West standard errors using the instruction in this post https://support.sas.com/kb/40/098.html

However, my dataset is quite large so SAS returns a Out of memory error. Is there a way to achieve this task for a large regression with many independent variables?

Here is a copy of the error:

ERROR: Java virtual machine exception. java.lang.OutOfMemoryError: GC overhead limit exceeded.
ERROR: Java virtual machine exception. java.lang.OutOfMemoryError: GC overhead limit exceeded.

proc model data=Sample_reg_iv;
   instruments x1 x2 x3;
   r=b0+b1*x1 + b2*x2 + b3*x3;
   fit r / gmm kernel=(bart,7,0);
   ods output ParameterEstimates=temp7 ;
run;

here is my code :

Re: How to get Newey West standard errors for large a dataset/regression?

Rick_SAS — Sun, 13 Jun 2021 10:16:23 GMT

The error can occur when the ODS system attempts to make graphs of huge data. Use PLOTS=NONE:

proc model data=Sample_reg_iv PLOTS=NONE;

topic Re: How to get Newey West standard errors for large a dataset/regression? in Statistical Procedures

How to get Newey West standard errors for large a dataset/regression?

Re: How to get Newey West standard errors for large a dataset/regression?