[Martin Johanneson]

Hi Jeff, very nice to hear from you!
I am unfamiliar with that method. Is the method relevant if the coefficients from industry and country is relevant/mean something to me?

In my case, after discussing this above with Clyde, my only interest is that Industry and Country (in addition to Year) are accounted for, and I think I finally understood the concept that xtreg, fe is an appropriate method to address this
And also, I learned that firm-level is also important to adjust for, which xtreg, fe also takes into account

Best,
Martin

[Clyde Schechter]

Since you are doing this for a master's thesis, you should consult with your master's thesis advisor about what your institution's requirements in connection with this are. And you should do whatever is required. Different institutions, and even different departments in the same institution, may have different requirements about this. It might even be left to the discretion of your thesis advisor. So find out what is expected of you--your thesis advisor should be able to answer this question or point you to a resource that spells out what is required.

You may find that some of the requirements connected with this do not make much sense, or are even blatantly bad statistical practice. So make note of that, because later in your career you will have greater latitude to do things your own way, and you can do better then.

[Martin Johanneson]

I see. I will confer with my thesis advisor and let you guys know

Thank you again


By the way. I ran the Hausman test which confirmed that FE is an appropriate model

Best,
Martin

[Martin Johanneson]

Hi again

I talked to my thesis-advisor and it turned out that OLS regression is be the appropriate model in my case. Apparently, the main goal is to investigate the relationships between X and Y across firms in the panel and not within firms. Therefore, the OLS regression with reg Y X covariates i.industry i.year i country

New topic :

1) In regard. How should I test for heteroskedasticity with the Breuch Pagan test?

First I run the OLS regression

2) Then which command do I have to use, is it xttest0 or the hettest command?

3) After discovering heteroskedasticity is evident, which vce should one use? The vce rob or vce (cluster firmid) in regard to investigating the relationship between X and Y across firms across year and not within?

Thank you in advance
Best
Martin