dropping constant in a fixed effects model

[Bridget Kauma]

Dear Statalist members,

I just wanted to check if there is a way I can specify a fixed effects model without a constant in Stata. When I run my regression it is dropping the first observation. Is there a way I can recover coefficients of the dropped observation.

Best,

Bridget

[Carlo Lazzaro]

Bridget:
without more quantitative data, it's difficult (for me, at least) to reply positively.
My guess is that your post includes two different questions:
1) the constant in -fe- (-xtreg-, I assume), is negligible (see: https://www.stata.com/support/faqs/s...effects-model/).
2) by default Stata drops observations with missing vakues in any variables. That said, if you wrote -observation- but you meant -variable- (that makes sense considering you mentioned coefficient), please note that the -fe- machinery wipes-out time-invariant predictors.

[Phil Bromiley]

The constant in a fe model is just the intercept for whichever panel doesn't get a fixed effect, so it is meaningless - change the dropped panel and you change the constant.