是通过蒙特卡罗模拟得到的结果么？

俺也想知道，搬凳子坐下先！

我认为是这个原因:

1. When conducting OLS or GLS, The coefficient value or significant level may be serious upward biased due to failures of some assumptions , like collinearity, E(U)`=0, autocorrelation striction, or even efficient issues

2. FE or RE may have downward result: FE lose too many degree of freedom, RE will be downward biased if E[u|X]=0.

3. GMM is a good estimator to deal with autocorrelation and heterogeneity issues.

讲得不对的方面,请达人指教.

所以GMM伟大啊