I am using reg3 command to estimate simultaneous equations, it looks the same instrumental variables are used in all equations under thid command . But in my simultaneous equations, some control variables are endogenous, so I need use different instruments for different equations, actually I should use GMM 3SLS estimator according to Wooldridge(2010).But I don't know how to implement this estimation procedure. Can anybody give me some tips? Thanks a lot.

George Gao