Weak identification test

[David Crawford]

Hello,

I am doing an estimation which involves running a IV regression with 7 endogenous variables and 8 instruments. I am using the Stata command xtivreg2 and it reports a Kleibergen-Paap Wald rk F statistic equal to 8.65. I would like to compare this value to the Stock-Yogo weak ID test critical values for 7 endogenous variables (K=7) and 8 instruments (L=8). However, such value is not reported by Stata, which, instead only provides the F test critical values for single endogenous regressor.

I would like to ask you whether Stata has a way of providing the Stock-Yogo critical values for 7 endogenous variables and 8 instruments for the different level of bias as well.

Thank you in advance for any help and suggestion

[George Ford]

are you certain that the test is not incorporating that fact (aside from what the output says)?

[George Ford]

https://www.stata.com/statalist/archive/2014-01/msg00483.html

[David Crawford]

Hi,

Thank you for the reply. I am not entirely sure that it is incorporating this. There is nothing on the documentation which seems to suggest otherwise. Do you know of the existence of another package perhaps does provide it?

Best

[George Ford]

There's an email address in the help file. Contact the author.

Given the nature of the test, I'm pretty sure it is doing what you seek.