How does one check the validity of instruments for a just-identified model, since the Sargan test can only be performed if the model is overidentified?
$\begingroup$
$\endgroup$
asked Mar 30, 2016 at 16:54
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
In a just-identified model you cannot statistically test the validity of your instruments. You are forced to rely on your subject-matter judgment regarding the exclusion restriction. As you point out, the Sargan test statistic is exactly 0 under just-identification, so useless as a test statistic.
The situation is in some sense similar to OLS - OLS is such that residuals and regressors are by construction orthogonal to each other (i.e., $X'\hat{u}=0$), so you cannot use that lack of correlation to argue that the assumption required for causal interpretation of OLS coefficients, namely $E(X'u)=0$, does hold.
answered Apr 6, 2016 at 8:38