I want to regress on only fixed effects to obtain dipersion of within-variation. However, I am using high-dimensional FEs, which means it is not feasible to do
Code:
reg y i.x1
because the number of different x1 is too large.

How can I achieve it?