I am reading a paper and try to implement its methodology in Stata. The paper says the following:
I understand the idea but not sure how to implement this in Stata. Stata's GMM estimator seems to use the condition E(Zu) = 0, where Z are instruments and u are regression residuals. I am not sure how to transform the sample moments are a function of unknown parameters to the E(Zu)=0 condition.
Equations (1)-(3) depend on only seven parameters. At the same time, we have seven sample means that can be calculated from the data. The expected values of these sample moments are a function of seven unknown parameters. Therefore, they provide a basis for a GMM estimator that is just-identified.