Dear all,
I am running a 2sls IV estimator using ivreg2. Initially, the used instrumental variables perform well and passed all tests in stata. However, when I add one more variable - the regional dummy - to the list of control variables, the instrumental variables do not pass tests on weak instruments. Although the regional dummy shows no significant effect on the outcome using IV (with the presense of weak IV issue), it is significant different from zero in OLS. Thus, excluding the regonial dummy in the equation may suffers ommitted varaible bias. However, since IV estimator address the issue of omitted variable, may I just exclude the regional dummy from the control variables please?
By the way, the used instrumental variables are all insignificant if added in OLS estimator.
Looking forward to your answers and suggestions.
Best,
Ziming
I am running a 2sls IV estimator using ivreg2. Initially, the used instrumental variables perform well and passed all tests in stata. However, when I add one more variable - the regional dummy - to the list of control variables, the instrumental variables do not pass tests on weak instruments. Although the regional dummy shows no significant effect on the outcome using IV (with the presense of weak IV issue), it is significant different from zero in OLS. Thus, excluding the regonial dummy in the equation may suffers ommitted varaible bias. However, since IV estimator address the issue of omitted variable, may I just exclude the regional dummy from the control variables please?
By the way, the used instrumental variables are all insignificant if added in OLS estimator.
Looking forward to your answers and suggestions.
Best,
Ziming
