In a multiple instrument regression using ivreg2 with gmm2s option I'm using partial. Despite using partial option, I get the following error which I attached at the last.
My dependent variable is a dummy variable, endogenous variable is a dummy variable. The multiple IV is the interaction of dummy variable of year and dummy variable of another variable - which takes up to 18 dummy values.
When I'm keeping the IV as i.year#c.dummyb then the regression runs and gives me Hansen J statistics. But when I'm using this as IV i.year#i.dummyb then I receive the following error despite keeping everything else the same.
This error I got after the regression running on my laptop for 2.5 hours. So, before experimenting anything else I would look for some kind guidance. Please help if you are familiar with this.
Code:
ivreg2 dep_var (endogenous_var= i.year#i.dummyb) male ismarried wasmarried age age2 black asian hispanic lths hsdegree somecollege i.year i.county $SCONTROL , gmm2s robust cluster(county) partial(i.county)
When I'm keeping the IV as i.year#c.dummyb then the regression runs and gives me Hansen J statistics. But when I'm using this as IV i.year#i.dummyb then I receive the following error despite keeping everything else the same.
Code:
Error: estimated covariance matrix of moment conditions not of full rank,
and optimal GMM weighting matrix not unique.
Possible causes:
singleton dummy variable (dummy with one 1 and N-1 0s or vice versa)
partial option may address problem.
r(506);
end of do-file
r(506);
