proc hpmixed with fixed effects and clustered standard error

[HEB1]

hello

I run this code: 

proc hpmixed data= mydata;
class fyear two_digit_SICnum FIRMid;
model ROA_NI = post__treatment  log_ta fyear two_digit_SICnum FIRMid /solution;
run;
quit;

i want to add to this code clustered stanard error by FIRMid. how can I do it? 

[sbxkoenk]

Hello,

I believe you have panel data (time series cross-sectional data) , right?

• fyear being the time identifier and
• FIRMid defining the cross-sections

I would use PROC PANEL or PROC CPANEL (CAS-equivalent) to analyze your data. You are in the econometrics space here.

And here's how to do the cluster correction of standard errors :

Usage Note 67322: Heteroscedasticity and cluster correction of standard errors using the PANEL procedure
https://support.sas.com/kb/67/322.html

Koen

[sbxkoenk · Accepted solution]

On top of my previous reply ...

If you want to stick to PROC MIXED or PROC HPMIXED, then look at the below topics.
They both contain the solution for what you want to achieve.

• Home > Analytics > Statistical Procedures >
  Cluster errors in Proc GLM
  https://communities.sas.com/t5/Statistical-Procedures/Cluster-errors-in-Proc-GLM/td-p/861830
• Home > Analytics > Statistical Procedures >
  Fixed effects with cluster
  https://communities.sas.com/t5/Statistical-Procedures/Fixed-effects-with-cluster/td-p/428211

proc mixed empirical;
  class firm;
  model y = x1 x2 x3 / solution;
  random int / subject=firm;
run;

Koen