Hi all,
I would like to ask you an econometric question concerning a panel dataset with fixed effects at the individual level in which I need to conduct some heterogeneity analysis.
Consider 3 variables:
Y, continuous
X, continuous
Z, categorical dummy (e.g. could be equal to 1 if democrat and 0 otherwise)
Consider also
X1 = X*Z, which is 0 when Z=0
X2 = X*(1-Z), which is 0 when Z=1
In order to estimate heterogenous effects of X on Y based on Z I would like to estimate a model of the form:
Y = b1X1 + (b1+b2)X2. Sample: all observations
All other techniques I tried don't work (e.g. sample split based on "if" condition).
The problem is that my supervisor said that I can't estimate the previous model keeping all the n individual fixed effects at the same time, and that Stata will automatically drop one.
I don't see why that should happen, because I am not convinced that the estimation gives rise to multicollinearity concerns. Is it that the case instead?
Many thanks in advance for your help and support!
I would like to ask you an econometric question concerning a panel dataset with fixed effects at the individual level in which I need to conduct some heterogeneity analysis.
Consider 3 variables:
Y, continuous
X, continuous
Z, categorical dummy (e.g. could be equal to 1 if democrat and 0 otherwise)
Consider also
X1 = X*Z, which is 0 when Z=0
X2 = X*(1-Z), which is 0 when Z=1
In order to estimate heterogenous effects of X on Y based on Z I would like to estimate a model of the form:
Y = b1X1 + (b1+b2)X2. Sample: all observations
All other techniques I tried don't work (e.g. sample split based on "if" condition).
The problem is that my supervisor said that I can't estimate the previous model keeping all the n individual fixed effects at the same time, and that Stata will automatically drop one.
I don't see why that should happen, because I am not convinced that the estimation gives rise to multicollinearity concerns. Is it that the case instead?
Many thanks in advance for your help and support!
Comment