Thanks Carlo Lazzaro !
In my case i am just trying for first-step OLS to check whether instruments used have any explanatory power on my endogeneous regressor (the insrument relevance assumption). Since instruments are time constant, so they will be dropped in fe estimator - That is why fe is not viable here, for the same reason i tried re estimator that gives me coefficeint with significance level on time constant instruments to test (crudely) for the instrumnet relevance, but re reports wald not F-statistics, so my question was- can we report wald, that instruments jointly determine endogeneous regressor after controlling for other covariates? My main estimator is ivpoisson with gmm , but before that, i want to make sure about the instrument relevance used further in my baselin estimator (ivpoisson).
Apart from that, the ivregress allows to test for weak instrument problem using post-estimation estat firststage which reports F(#,#). I was wondering, whether there is any such post-estimation procedure to get F-statistics in re effect estimator using xtreg command - This is other way of asking my question. I hope this helps in understanding.
Thanks,
(Ridwan)
In my case i am just trying for first-step OLS to check whether instruments used have any explanatory power on my endogeneous regressor (the insrument relevance assumption). Since instruments are time constant, so they will be dropped in fe estimator - That is why fe is not viable here, for the same reason i tried re estimator that gives me coefficeint with significance level on time constant instruments to test (crudely) for the instrumnet relevance, but re reports wald not F-statistics, so my question was- can we report wald, that instruments jointly determine endogeneous regressor after controlling for other covariates? My main estimator is ivpoisson with gmm , but before that, i want to make sure about the instrument relevance used further in my baselin estimator (ivpoisson).
Apart from that, the ivregress allows to test for weak instrument problem using post-estimation estat firststage which reports F(#,#). I was wondering, whether there is any such post-estimation procedure to get F-statistics in re effect estimator using xtreg command - This is other way of asking my question. I hope this helps in understanding.
Thanks,
(Ridwan)
Comment