Hello,

I'm somewhat puzzled that stata is able to estimate an interaction term of a variable that does not vary within the fixed effect group.

Suppose I have the following code

Code:
reg lhs c.rhs_g##c.x id.i
where lhs and rhs_g vary across id and time and x only varies across time, but is constant within id. I would have expected a multi-collinearity problem where stata cannot estimate the interaction term. Why is this reasoning wrong?