Dear all,
I ran two simultaneous, non linear, equations using -gmm- command (Stata 14.1).
I have 2 endogenous var. in each equation, and 6 IV's for these two variables (overidentification). The -estat overid- test (Hansen's J statistic) indicates that there is no problem of validity the overidentifying restrictions. Does that mean that all IV's are orthogonal to the errors?
Second, how do I test whether the IV's are correlated with the included endogenous variable? since the first step results are not available in the -gmm- command, and -ivendog- is also not available, I am not sure what procedure can I use, other then running individual correlation tests (between the endogenous regressors and the IV's)
Thanks.
Anat
I ran two simultaneous, non linear, equations using -gmm- command (Stata 14.1).
I have 2 endogenous var. in each equation, and 6 IV's for these two variables (overidentification). The -estat overid- test (Hansen's J statistic) indicates that there is no problem of validity the overidentifying restrictions. Does that mean that all IV's are orthogonal to the errors?
Second, how do I test whether the IV's are correlated with the included endogenous variable? since the first step results are not available in the -gmm- command, and -ivendog- is also not available, I am not sure what procedure can I use, other then running individual correlation tests (between the endogenous regressors and the IV's)
Thanks.
Anat