anyone familiar with GMM in stata? I'm currently attempting to replicate a methodology utilized in a Rigobon(2002) paper, in which he says "a VAR is run on the log of the yields to remove effects of serial correlation...Once the subsamples are defined, teh covariance matrix in each of them is computed. Those covariance matrices are used in the GMM estimation of the contemporeaneous coefficients. The standard errors are computed by bootstrapping. The residuals are bootstrapped to obtain a distribution of covariance matrices."

So - I believe I have the VAR part down. How do I apply GMM to the covariance matrices, and how do I bootstrap the standard errors? I've literally looked for help all over the Internet, and the one GMM powerpoint proved unhelpful since I kept on getting error messages.