Proc glm for panel data

[msj]

Hi everybody

I am using an unbalanced panel data set over individuals and want to estimate pooled ols(POLS), fixed effect (FE) and random effect (RE). Furthermore, I would like to make a hausman test, in order to see which of the FE or RE is preferred.

I cannot use the proc panel command, as my dataset is too big (error says insufficient memory). So far I have used proc glm, but I am not sure about its ability to produce robust standard errors from and it does not seem possible to perform a hausman test using this statement. Are there any better solutions for these estimations?

My current code looks like this:

POLS

Proc Glm data=merged;
Class year edu municipal;
Model Lnw= age lnedens lnfsize edu year municipal /solution;
Run;

FE

Proc Glm data=merged;
Absorb id;
Class year edu municipal;
Model Lnw= age lnedens lnfsize edu year municipal /solution;
Run;

RE

Proc Glm data=merged;
Class year edu municipal;
Model Lnw= age lnedens lnfsize edu year municipal /solution;
Random age lnedens lnfsize edu year municipal;
Run;

The RE gives me the same as POLS, but with the error statement saying only class variables are allowed in random statement. When I try this, I do not get any estimates.

Any suggestions as to how I can perform these estimations in a better way?

Thank you

[bobby_sas · Accepted solution]

Because you would like a Hausman test, you could try the following to get PROC PANEL to work.

1. Increase the memory footprint so that the dataset may fit.  See the following link for the MEMSIZE system option:

http://support.sas.com/documentation/cdl/en/hostwin/69955/HTML/default/viewer.htm#n0srij4mcdpzv2n1vt...

2. Consider using the high-performance version of PROC PANEL: PROC HPPANEL.  See

http://support.sas.com/documentation/cdl/en/etshpug/68149/HTML/default/viewer.htm#etshpug_hppanel_to...

Finally, we continuously make performance improvements to our panel-data procedures in SAS/ETS.  If you would like to email me at Bobby.Gutierrez@sas.com and let me know the version of SAS/ETS you are using, and how big of a dataset we're dealing with, I'd be glad to provide more specific advice.

--bobby