Hello,
I have a panel dataset and am running -IVREG2H-. I have 2 endogenous regressors. I am only relying on generated instruments. The equation looks like the following where Y is dependent variable and X4 and X5 are probably endogenous.
ivreg2h Y X1 X2 X3 (X4 X5=), cluster(firm years)
Question 1) The -ivreg2h command output includes 3 different statistics. One of which is "Underidentification test (Kleibergen-Paap rk LM statistic)". Is this statistic relevant if I only have generated instruments no excluded instruments? I want to confirm if I am right.
Question 2) I want to know the difference between the following 2 approaches if any -
Option 1: Include both endogenous regressors together in a single equation.
ivreg2h Y X1 X2 X3 (X4 X5=), cluster(firm years)
Option 2: Test for the 2 endogenous regressors separately.
ivreg2h Y X1 X2 X3 (X4=), cluster(firm years)
ivreg2h Y X1 X2 X3 (X5=), cluster(firm years)
Thanks.
I have a panel dataset and am running -IVREG2H-. I have 2 endogenous regressors. I am only relying on generated instruments. The equation looks like the following where Y is dependent variable and X4 and X5 are probably endogenous.
ivreg2h Y X1 X2 X3 (X4 X5=), cluster(firm years)
Question 1) The -ivreg2h command output includes 3 different statistics. One of which is "Underidentification test (Kleibergen-Paap rk LM statistic)". Is this statistic relevant if I only have generated instruments no excluded instruments? I want to confirm if I am right.
Question 2) I want to know the difference between the following 2 approaches if any -
Option 1: Include both endogenous regressors together in a single equation.
ivreg2h Y X1 X2 X3 (X4 X5=), cluster(firm years)
Option 2: Test for the 2 endogenous regressors separately.
ivreg2h Y X1 X2 X3 (X4=), cluster(firm years)
ivreg2h Y X1 X2 X3 (X5=), cluster(firm years)
Thanks.