interpreting IV diagnostics and tests

[mattemde]

I am using Eviews 7. I am doing GMM estimation and want to use the IV diagnostics and tests that Eviews has available. But I am having difficulty interpreting the test results.

For the regressor endogeneity test, is the null hypothesis that the regressors are endogenous (for example, the p-value I'm getting is 0.80, which would mean the regressos are endogenous)?

For the weak instruments diagnostics, is the null hypothesis for the Cragg-Donald F-statistic that the instruments are valid? And for the MSC, are lower numbers better when comparing between competing instrument sets?

Thanks for your help,
Matt

[EViews Glenn]

For the Endogeneity test, the null is that the instrument subset is exogenous.

For the Stock-Yogo variant of the Cragg-Donald statistic the null is that the instruments are weak, even though the parameters might be identified.

[mattemde]

For the Endogeneity test, the null is that the instrument subset is exogenous.

I don't understand why the null refers to the instrument subset rather than the regressor subset. Are you referring to the Instrument Orthogonality Test, perhaps? For the Regressor Endogeneity Test, I thought the purpose was to determine whether a regressor is endogenous (explained by the instruments) or exogenous (not explained by the instruments).

Thanks,
Matt

[EViews Glenn]

Sorry, I was describing the instrument orthogonality test...

The standard Durbin-Wu-Hausman null is that the candidate variables are exogenous (and are therefore valid instruments).

[haji]

Hi,
Kindly help me out as to how to interpret results for GMM. I got the following results:
for instrument orthogonality test: Difference in J-stats -1.33E-15 0 NA: i do not understand why df is zero
for endogeneity test: Difference in J-stats 0.003204 1 0.9549: what is null hypothesis in it...
Weak Instrument Diagnostics:
Cragg-Donald F-statistic: 0.434938
Stock-Yugo critical values (relative bias):
5% 19.28
10% 11.12
20% 6.76
30% 5.15
Stock-Yugo critical values (size):
10% 29.18
15% 16.23
20% 11.72
25% 9.38
how do we interpret cragg-donald.....

[ronanuga]

Dear all,
Please help me with the interpretation of "Weak Instrument Diagnostic test".
I know the null hypothesis is that the instruments are weak. The result is:
Cragg-Donald F-stat: 0.034220
Stock-Yogo critical values (size):
10% 13.43
15% 8.18
20% 6.40
25% 5.45
Your response is appreciated, thank you.