No constant in fixed-effect regression

[Long Hong]

Dear all:

Hope you are doing well and healthy. I am writing to ask a quick question about the fixed effect regressions in Stata.

I notice that none of the fixed-effect commands in Stata (areg, xtreg, reghdfe) could suppress the constant term. However, the constant term is quite annoying if my key interest is the fixed effect estimates because it could mess up the fixed effect estimates.

I am wondering if anyone knows any smart way to get rid of the constant term or recover the true fixed-effects estimates?

My hunch is:

xtreg y x, fe
predict fe_wrong, u
gen fe_right = fe_wrong + _b[_cons]

But I am not 100% sure.

If it is the case, how could we recover the true two-way fixed effects from constant?

Thank you, and I look forward to hearing from you!

Best,
Long

[Joseph Coveney]

My ignorance chagrins me, but is there a reference that defines what you mean by the true fixed-effects estimate? And explains how the intercept messes it up?

[Carlo Lazzaro]

Long:
have you already taken a look at: https://www.statalist.org/forums/for...stored-results ?

[Arthur Morris]

I'm not sure I understand everything else you're asking, but xtreg, fe does not report the constant you think it does. The constant reported is the average of the fixed effects.
Recommended reading: How can there be an intercept in the fixed-effects model estimated by xtreg, fe?

The documentation of reghdfe explains how to handle the constant it produces.