Dear Statalist users,
I am trying to do a two-step system gmm estimation for a growth model (how gini affects growth, besides other control vars) but I have a problem with missing data. For one of my variable I only have data for every 5 years and due to this, almost all of my other years are dropped from the regression due to collinearity. Since the variable is important, I cannot just drop it.
A solution may be to impute the missing data for the missing years (the variable doesn't have a high variance within groups so I think it wouldn't hurt the efficiency of the regression). But does Xtabond2 supports the "mi estimate prefix"? I saw it in older posts that it won't produce the post-estimation (AR tests and the Hansen-Sargan tests) but those forums were open 9 years ago.
Thank you for the help in advance!
I am trying to do a two-step system gmm estimation for a growth model (how gini affects growth, besides other control vars) but I have a problem with missing data. For one of my variable I only have data for every 5 years and due to this, almost all of my other years are dropped from the regression due to collinearity. Since the variable is important, I cannot just drop it.
A solution may be to impute the missing data for the missing years (the variable doesn't have a high variance within groups so I think it wouldn't hurt the efficiency of the regression). But does Xtabond2 supports the "mi estimate prefix"? I saw it in older posts that it won't produce the post-estimation (AR tests and the Hansen-Sargan tests) but those forums were open 9 years ago.
Thank you for the help in advance!
Comment