Dear users,
I have a dynamic panel dataset that I would like to analyze using xtabond2, but I'm having a hard time deciding which of my right hand side variables should be declared endogenous and which ones shouldn't. The problem is that a priori, I can think of reasons (theoretically) as to why several variables should be declared endogenous. And yet when I proceed to do that, I have problems making my models converge. Instead, I would like to use the criterion that "the variable is correlated with past of present realizations of the error term" (Roodman 2009). Is there any way to test for this in Stata? I have searched online but could not find an answer to my question.
Thank you,
I have a dynamic panel dataset that I would like to analyze using xtabond2, but I'm having a hard time deciding which of my right hand side variables should be declared endogenous and which ones shouldn't. The problem is that a priori, I can think of reasons (theoretically) as to why several variables should be declared endogenous. And yet when I proceed to do that, I have problems making my models converge. Instead, I would like to use the criterion that "the variable is correlated with past of present realizations of the error term" (Roodman 2009). Is there any way to test for this in Stata? I have searched online but could not find an answer to my question.
Thank you,
Comment