I get low and negative R-squares. Is R-squared not important as the coefficient of determination in GMM? I know R-squared is important in the first equation of 2SLS, but if I am estimating a pure GMM, I get low and even negative R-squares. How is this interpreted, or it is not important?
I got a PhD thesis but in Arabic language. Is it useful for you?
Are you need to send reference of the above thesis in English language?
http://forums.eviews.com/viewtopic.php?f=4&t=2729
It seems that R squared no longer relfect the fit of the model, so you need to be sure about the validity of the instruments only, through Sargan test or J-statistic
good luck
Dear Temitope Lydia Leshoro
As mentioned by Mr. Said, in these robust models like GMM R-Square is not the point to bother. If you find valid instruments, producing good results and also free from the over/underestimation and weak instrument hypothesis, you can carry on with results.
Dear Said Jaouadi,
Thanks for the answer, can you provide some links for reference literature.
Based on theory and prior literature, R Squared is not important in both System and Difference GMM.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude