[R] package : plm : pgmm question

[Ye Dong]

Dear Sir,



I am using the package pgmm you build in panel regression. However, I found that when T is 10, N=30, the error would show as following:



system is computationally singular: reciprocal condition number

But the similar code works well on Stata, so I wonder  how I can optimize the algorithm, for example , the inverse matrix optimization ? And I have checked my data as well, no multicollinearity problem exists. Another problem is that although I have some NA in the panel data, the panel dataframe is still recognized as balanced model. But with plm, the dataframe would be recognized unbalanced.

Thanks and Best regards,

Ye Dong


	[[alternative HTML version deleted]]