Dear Statalisters,

I am new to the forum and I have tried to find a solution for my problem. I could not find it but please let me know if another forum already covers it.

To put things simply, let's say I have the following regression:

xtreg y_it X1_it + X2_i c.X1l##c.X2 with X1 and X2 being continuous variables.

If I run a panel data regression with fixed effects on i, the coefficient related to X2 wipes out.

Yet, if I want to look at the marginal effect of X1_it on y_it it should be: b1 + b2X2_i.

That said, after running my panel data regression and calling:

margins, dydx(X1) at(X2=(min(intervall)max)) - of course replacing for relevant numbers and variables in my regression.

Yet, it gives me a "(not estimable output)" - see image below. If I run the regression with random effects, the command works.

Thus, I believe it has to do with the fact that the X2 variable is time invariant although theoretically I do not understand why I can't get the marginal effects for different levels of variable X2.

Thank you in advance for your help.

Kind regards,

Antoine