Dummy variables in fixed effect panel regression

[Ugne Enchte]

Hello,

I have some panel data with the following variables:
ID=1,2,3,...
Year: 2010-2020
ROA: ratio from -2 to 1
GEAR: ratio from 0 to 8
OWN: dummy variable showing if ownership is foreign or domestic with the values 0 and 1
other control variables.

I define panel as: xtset ID YEAR, yearly.

When I run: xtreg ROA GEAR SIZE AGE ib0.OWN, fe vce (robust)

it omits OWN. When I do a correlation matrix or regress OWN with other variables there is no correlation and R2 is nowhere near 1.

Why does Stata exclude my dummy variable and how else can I incorporate the effect of foreign ownership on my results?

Any help is appreciated!

[Carlo Lazzaro]

Ugne:
welcome to this forum.
If -OWN- is a time- invariant variable, as expected it is wiped out by the -fe- machinery.

[Ugne Enchte]

Hello Calro, thank you!

OWN is time invariant and only changes by the entity. Can you please suggest any code that I can use to see the effect of OWN as a control variable when regressing GEAR on ROA, while also including my other variables? I used xtreg and fixed effect up to now.

[Carlo Lazzaro]

Ugne:
if you want to go -fe- there's no chance that you can estimate that coefficient.
Options to reach that goal are -xtreg,re. and the community-contributed module -mundlak-.
As -fe-, -re- (and -mundlak-) are not interchangeable and give their best under a set of conditions, you should check whether they're satisfied.