Hi,
I have a model with 3 endogenous variables and 4 instruments. The individual F-stats for first stage regressions are high (>50), however Cragg-Donald statistics is <7.
My questions are:
1. Why would Cragg-Donald statistic be so low when individual first stages have high F-stats?
2. Is there an alternative test that could be used in such scenario of more than one endogenous variables?
3. Is it possible that a high correlation between endogenous variables could cause this?
Thank you
Parth
I have a model with 3 endogenous variables and 4 instruments. The individual F-stats for first stage regressions are high (>50), however Cragg-Donald statistics is <7.
My questions are:
1. Why would Cragg-Donald statistic be so low when individual first stages have high F-stats?
2. Is there an alternative test that could be used in such scenario of more than one endogenous variables?
3. Is it possible that a high correlation between endogenous variables could cause this?
Thank you
Parth
