Dear Statalisters!
I'm working with panel dataset in stata. As long as I have 35 companies and 84 monthly observations my dataset is "T>N" type.
I know it is ok to apply FE, RE with Driscoll-Kraay standard errors (-xtscc-) or PCSE (-xtpcse-) when T>N.
I've tested the standard FE and RE models with Driscoll-Kraay standard errors using Mundlak test:
https://blog.stata.com/2015/10/29/fi...dlak-approach/
Result of the Mundlak test suggests I should add fixed effects in the model.
At the same time I would also check the results with PCSE model as It's highly recommended in T>N environment.
However the fixed effect option isn't included in standard -xtpcse- command.
Whether it is ok to add company fixed effects in the standard PCSE model like that:
?
Best regards, Pavel.
I'm working with panel dataset in stata. As long as I have 35 companies and 84 monthly observations my dataset is "T>N" type.
I know it is ok to apply FE, RE with Driscoll-Kraay standard errors (-xtscc-) or PCSE (-xtpcse-) when T>N.
I've tested the standard FE and RE models with Driscoll-Kraay standard errors using Mundlak test:
https://blog.stata.com/2015/10/29/fi...dlak-approach/
Result of the Mundlak test suggests I should add fixed effects in the model.
At the same time I would also check the results with PCSE model as It's highly recommended in T>N environment.
However the fixed effect option isn't included in standard -xtpcse- command.
Whether it is ok to add company fixed effects in the standard PCSE model like that:
Code:
xtpcse y x1 x2 x3 i.CompanyName_1, corr(psar1)
Best regards, Pavel.
