dear all i have a very urgent question and I dont know what to do anymore. my proposal has been rejected and i only have 2 days to change it..
basically i ran an OLS regression and included an interaction term between a time invariant and time variant variable, i included the MAIN variable aswell and firm fixed effets. now I GOT COEFFICIENT ESTIMATES ON THE MAIN VARIABLE, THE INTERACTION TERM WHILE I INCLUDED FIRM FIXED EFFECTS. This is not supposed to happen as the main time invariant variable is supposed to drop out.
I re ran every possible specification but this keeps happening. Can you please help me
I used the xi: reg x1 x1*x2 x2 i.year i.firm command
thanks a lot
What's your outcome variable here? Why have you interacted it with a predictor variable?
You should be using -xtreg- for a fixed-effects regression. Assuming you just forgot to include "y" in your command above, and assuming that x1 and x2 are continuous variables, try something like:
Code:xtset firm year xtreg y c.x1##c.x2 i.year, fe
Hi,
Thank you very much for your reply. Basically my outcome variable is credit growth (y = increase in debt). My main indep. var is firms engaged in R&D (dummy variable). X1 = thus dummy and x2 = continous. X2 = profit margin, interaction is x1*x2. I searched for how to use interaction terms and i saw that one must include the main independent var so that there is no omitted variable bias. When i run the regression (also using xtreg I tried that too) the coefficients on my main independent variable are calculated and they are significant. This is not suppose to happen because I included firm fixed effects and my main indep. var is firms engaged in R&D and time invariant. I tried everything every possible command but the variable is not dropped.
Can you run the code I posted above please, and show the command and output? Also please post the result of -xtsum-
Posting Permissions
|
|
Reply With Quote