I have been asked by reviewer to conduct model misspecification test while i am using system GMM technique. I never saw such sort of test in any paper where GMM is used. Guidance is requested, please.
Muhammad Kamal , I think you should test for over-identification restrictions and serial correlation in dynamic panel models. Arellano and Bond (1991) suggested the Sargen test for over-identifying restrictions. The test can be obtained with stata command estat sargen. Other test suggested is the Sargan's difference static to test for nested hypotheses concerning serial correlation.