Pooled cross-sections with instruments-HELP

[Stavroula Sarri]

Hello !!

It's the first time I am using pooled cross-sectional data and I am a bit confused. I need to use instruments in order to overcome any endogeneity problem that might arise.
My sample constitutes of 2,670 firm- observations (T=10 -consecutive years- & N= 267). I have run ivreg2 with liml, but still I am not sure if this is the appropriate model.
My model passes successfully al the diagnostic tests for instruments validity, weakness and power.

My command is:

ivreg2 sa1 (zs=degree_m mne_yn duration i_seniority reliability) cr1 size1 prof1 if sample1==1, liml robust

cr1,size1 and prof1 are the control variables.

Do I have to include year or country dummies?

Thank you very much for your time and I am sorry for the elementary question.

Best Regards,
Stavroula

[Phil Bromiley]

You're almost certainly better off with xtreg or xtreg2. You probably want some panel controls but whether you want firm, industry, or country is a matter of substance. In some research areas, folks automatically assume panel effects at the firm level. If you include firm, then you don't need more aggregate effects like industry or country since firms don't change industry or country and so take care of them.

Likewise, whether you include year fixed effects depends on the substance of your theory and model.