Residual Normality and GMM

[learner]

Dear Sir

I want to know, what is the significance of Residual Normality in GMM, Is it as important as in OLS or more or less?

Thanks

Learner

[wolly77]

Hi,

I am not an expert on GMM estimations, but from the paper that I have read (Phillips curve estimation) it is clear that normality test is never reported. In estimations HAC weighting matrix is used for correcting for serial correlation and heteroskedasticity, and stop. In some cases (i.e., most recent papers) weak instruments tests are also used (Anderson and Rubin (1949)). But residual normality is never tested.