How to do Fixed Effects on my own in Stata

[Eren Tina]

Dear experts,
I learned from my lecturer that we'd better estimate FE estimators by using the existing programs, like reg i.id, xtreg, instead of doing it on our own. And I tried to do it on my own and compare my result to the existing commands. It's not the same.
I also learned that my errors come from the degrees of freedom, but I don't know how to adjust it mannually.
Can you teach me how to adjust it? Thank you so much.

the reghdfe result:

the reg result: I wonder how I can adjust the t-statistic to make it right

[Eren Tina]

I rewrite my command like this：reg Tobin_Q GINDEX Size Leverage ROA, vce(cluster Gvkey)
according to Computing correct standard errors for within estimator - Statalist
but the t statistic is still wrong

[FernandoRios]

Hi Eren
that isnt enough.
you need to consider that in the original model (one with fixed effects included explicitly, you had other 3000's estimated parameters (based on your output)
When you estimate the demean model, you are not counting those "implicit" parameters.
check this paper of mine where i explain how this works and how are STD ERR re estimated.
https://journals.sagepub.com/doi/pdf...867X1501500318
HTH

[Eren Tina]

Hi Eren
that isnt enough.
you need to consider that in the original model (one with fixed effects included explicitly, you had other 3000's estimated parameters (based on your output)
When you estimate the demean model, you are not counting those "implicit" parameters.
check this paper of mine where i explain how this works and how are STD ERR re estimated.
https://journals.sagepub.com/doi/pdf...867X1501500318
HTH

Thank you very much. It helps a lot